· 8 years ago · Nov 02, 2017, 10:56 PM
1commit 890da9cf098364b11a7f7f5c22fa652531624d03
2Author: Linus Torvalds <torvalds@linux-foundation.org>
3Date: Thu Nov 2 14:06:32 2017 -0700
4
5 Revert "x86: do not use cpufreq_quick_get() for /proc/cpuinfo "cpu MHz""
6
7 This reverts commit 51204e0639c49ada02fd823782ad673b6326d748.
8
9 There wasn't really any good reason for it, and people are complaining
10 (rightly) that it broke existing practice.
11
12 Cc: Len Brown <len.brown@intel.com>
13 Cc: Thomas Gleixner <tglx@linutronix.de>
14 Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
15 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16commit 80c094a47dd4ea63375e3f60b5e076064f16e857
17Author: Linus Torvalds <torvalds@linux-foundation.org>
18Date: Thu Oct 26 19:35:35 2017 +0200
19
20 Revert "apparmor: add base infastructure for socket mediation"
21
22 This reverts commit 651e28c5537abb39076d3949fb7618536f1d242e.
23
24 This caused a regression:
25 "The specific problem is that dnsmasq refuses to start on openSUSE Leap
26 42.2. The specific cause is that and attempt to open a PF_LOCAL socket
27 gets EACCES. This means that networking doesn't function on a system
28 with a 4.14-rc2 system."
29
30 Sadly, the developers involved seemed to be in denial for several weeks
31 about this, delaying the revert. This has not been a good release for
32 the security subsystem, and this area needs to change development
33 practices.
34
35 Reported-and-bisected-by: James Bottomley <James.Bottomley@hansenpartnership.com>
36 Tracked-by: Thorsten Leemhuis <regressions@leemhuis.info>
37 Cc: John Johansen <john.johansen@canonical.com>
38 Cc: Vlastimil Babka <vbabka@suse.cz>
39 Cc: Seth Arnold <seth.arnold@canonical.com>
40 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
41commit e49aa15ef6c179f69e5578a271801f31a09e9a3f
42Author: Linus Torvalds <torvalds@linux-foundation.org>
43Date: Thu Sep 28 13:20:32 2017 -0700
44
45 Revert "Bluetooth: Add option for disabling legacy ioctl interfaces"
46
47 This reverts commit dbbccdc4ced015cdd4051299bd87fbe0254ad351.
48
49 It turns out that the "legacy" users aren't so legacy at all, and that
50 turning off the legacy ioctl will break the current Qt bluetooth stack
51 for bluetooth LE devices that were released just a couple of months ago.
52
53 So it's simply not true that this was a legacy interface that hasn't
54 been needed and is only limited to old legacy BT devices. Because I
55 actually read Kconfig help messages, and actively try to turn off
56 features that I don't need, I turned the option off.
57
58 Then I spent _way_ too much time debugging BLE issues until I realized
59 that it wasn't the Qt and subsurface development that had broken one of
60 my dive computer BLE downloads, but simply my broken kernel config.
61
62 Maybe in a decade it will be true that this is a legacy interface. And
63 maybe with a better help-text and correct dependencies, this kind of
64 legacy removal might be acceptable. But as things are right now both
65 the commit message and the Kconfig help text were misleading, and the
66 Kconfig option had the wrong dependenencies.
67
68 There's no reason to keep that broken Kconfig option in the tree.
69
70 Cc: Marcel Holtmann <marcel@holtmann.org>
71 Cc: Johan Hedberg <johan.hedberg@intel.com>
72 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
73commit f007cad159e99fa2acd3b2e9364fbb32ad28b971
74Author: Linus Torvalds <torvalds@linux-foundation.org>
75Date: Sun Sep 10 21:19:06 2017 -0700
76
77 Revert "firmware: add sanity check on shutdown/suspend"
78
79 This reverts commit 81f95076281fdd3bc382e004ba1bce8e82fccbce.
80
81 It causes random failures of firmware loading at resume time (well,
82 random for me, it seems to be more reliable for others) because the
83 firmware disabling is not actually synchronous with any particular
84 resume event, and at least the btusb driver that uses a workqueue to
85 load the firmware at resume seems to occasionally hit the "firmware
86 loading is disabled" logic because the firmware loader hasn't gotten the
87 resume event yet.
88
89 Some kind of sanity check for not trying to load firmware when it's not
90 possible might be a good thing, but this commit was not it.
91
92 Greg seems to have silently suffered the same issue, and pointed to the
93 likely culprit, and Gabriel C verified the revert fixed it for him too.
94
95 Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
96 Pointed-at-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
97 Tested-by: Gabriel C <nix.or.die@gmail.com>
98 Cc: Luis R. Rodriguez <mcgrof@kernel.org>
99 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
100commit 785373b4c38719f4af6775845df6be1dfaea120f
101Author: Linus Torvalds <torvalds@linux-foundation.org>
102Date: Tue Aug 29 09:11:06 2017 -0700
103
104 Revert "rmap: do not call mmu_notifier_invalidate_page() under ptl"
105
106 This reverts commit aac2fea94f7a3df8ad1eeb477eb2643f81fd5393.
107
108 It turns out that that patch was complete and utter garbage, and broke
109 KVM, resulting in odd oopses.
110
111 Quoting Andrea Arcangeli:
112 "The aforementioned commit has 3 bugs.
113
114 1) mmu_notifier_invalidate_range cannot be used in replacement of
115 mmu_notifier_invalidate_range_start/end.
116
117 For KVM mmu_notifier_invalidate_range is a noop and rightfully so.
118
119 A MMU notifier implementation has to implement either
120 ->invalidate_range method or the invalidate_range_start/end
121 methods, not both. And if you implement invalidate_range_start/end
122 like KVM is forced to do, calling mmu_notifier_invalidate_range in
123 common code is a noop for KVM.
124
125 For those MMU notifiers that can get away only implementing
126 ->invalidate_range, the ->invalidate_range is implicitly called by
127 mmu_notifier_invalidate_range_end(). And only those secondary MMUs
128 that share the same pagetable with the primary MMU (like AMD
129 iommuv2) can get away only implementing ->invalidate_range.
130
131 So all cases (THP on/off) are broken right now.
132
133 To fix this is enough to replace mmu_notifier_invalidate_range with
134 mmu_notifier_invalidate_range_start;mmu_notifier_invalidate_range_end.
135 Either that or call multiple mmu_notifier_invalidate_page like
136 before.
137
138 2) address + (1UL << compound_order(page) is buggy, it should be
139 PAGE_SIZE << compound_order(page), it's bytes not pages, 2M not
140 512.
141
142 3) The whole invalidate_range thing was an attempt to call a single
143 invalidate while walking multiple 4k ptes that maps the same THP
144 (after a pmd virtual split without physical compound page THP
145 split).
146
147 It's unclear if the rmap_walk will always provide an address that
148 is 2M aligned as parameter to try_to_unmap_one, in presence of THP.
149 I think it needs also an address &= (PAGE_SIZE <<
150 compound_order(page)) - 1 to be safe"
151
152 In general, we should stop making excuses for horrible MMU notifier
153 users. It's much more important that the core VM is sane and safe, than
154 letting MMU notifiers sleep.
155
156 So if some MMU notifier is sleeping under a spinlock, we need to fix the
157 notifier, not try to make excuses for that garbage in the core VM.
158
159 Reported-and-tested-by: Bernhard Held <berny156@gmx.de>
160 Reported-and-tested-by: Adam Borowski <kilobyte@angband.pl>
161 Cc: Andrea Arcangeli <aarcange@redhat.com>
162 Cc: Radim KrÄmář <rkrcmar@redhat.com>
163 Cc: Wanpeng Li <kernellwp@gmail.com>
164 Cc: Paolo Bonzini <pbonzini@redhat.com>
165 Cc: Takashi Iwai <tiwai@suse.de>
166 Cc: Nadav Amit <nadav.amit@gmail.com>
167 Cc: Mike Galbraith <efault@gmx.de>
168 Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
169 Cc: Jérôme Glisse <jglisse@redhat.com>
170 Cc: axie <axie@amd.com>
171 Cc: Andrew Morton <akpm@linux-foundation.org>
172 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
173commit 143c97cc652949893c8056c679012f0aeccb80e5
174Author: Linus Torvalds <torvalds@linux-foundation.org>
175Date: Wed Aug 23 18:16:11 2017 -0700
176
177 Revert "pty: fix the cached path of the pty slave file descriptor in the master"
178
179 This reverts commit c8c03f1858331e85d397bacccd34ef409aae993c.
180
181 It turns out that while fixing the ptmx file descriptor to have the
182 correct 'struct path' to the associated slave pty is a really good
183 thing, it breaks some user space tools for a very annoying reason.
184
185 The problem is that /dev/ptmx and its associated slave pty (/dev/pts/X)
186 are on different mounts. That was what caused us to have the wrong path
187 in the first place (we would mix up the vfsmount of the 'ptmx' node,
188 with the dentry of the pty slave node), but it also means that now while
189 we use the right vfsmount, having the pty master open also keeps the pts
190 mount busy.
191
192 And it turn sout that that makes 'pbuilder' very unhappy, as noted by
193 Stefan Lippers-Hollmann:
194
195 "This patch introduces a regression for me when using pbuilder
196 0.228.7[2] (a helper to build Debian packages in a chroot and to
197 create and update its chroots) when trying to umount /dev/ptmx (inside
198 the chroot) on Debian/ unstable (full log and pbuilder configuration
199 file[3] attached).
200
201 [...]
202 Setting up build-essential (12.3) ...
203 Processing triggers for libc-bin (2.24-15) ...
204 I: unmounting dev/ptmx filesystem
205 W: Could not unmount dev/ptmx: umount: /var/cache/pbuilder/build/1340/dev/ptmx: target is busy
206 (In some cases useful info about processes that
207 use the device is found by lsof(8) or fuser(1).)"
208
209 apparently pbuilder tries to unmount the /dev/pts filesystem while still
210 holding at least one master node open, which is arguably not very nice,
211 but we don't break user space even when fixing other bugs.
212
213 So this commit has to be reverted.
214
215 I'll try to figure out a way to avoid caching the path to the slave pty
216 in the master pty. The only thing that actually wants that slave pty
217 path is the "TIOCGPTPEER" ioctl, and I think we could just recreate the
218 path at that time.
219
220 Reported-by: Stefan Lippers-Hollmann <s.l-h@gmx.de>
221 Cc: Eric W Biederman <ebiederm@xmission.com>
222 Cc: Christian Brauner <christian.brauner@canonical.com>
223 Cc: Al Viro <viro@zeniv.linux.org.uk>
224 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
225commit 558e8e27e73f53f8a512485be538b07115fe5f3c
226Author: Linus Torvalds <torvalds@linux-foundation.org>
227Date: Thu Feb 16 12:19:18 2017 -0800
228
229 Revert "nohz: Fix collision between tick and other hrtimers"
230
231 This reverts commit 24b91e360ef521a2808771633d76ebc68bd5604b and commit
232 7bdb59f1ad47 ("tick/nohz: Fix possible missing clock reprog after tick
233 soft restart") that depends on it,
234
235 Pavel reports that it causes occasional boot hangs for him that seem to
236 depend on just how the machine was booted. In particular, his machine
237 hangs at around the PCI fixups of the EHCI USB host controller, but only
238 hangs from cold boot, not from a warm boot.
239
240 Thomas Gleixner suspecs it's a CPU hotplug interaction, particularly
241 since Pavel also saw suspend/resume issues that seem to be related.
242 We're reverting for now while trying to figure out the root cause.
243
244 Reported-bisected-and-tested-by: Pavel Machek <pavel@ucw.cz>
245 Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
246 Cc: Wanpeng Li <wanpeng.li@hotmail.com>
247 Cc: Peter Zijlstra <peterz@infradead.org>
248 Cc: Rik van Riel <riel@redhat.com>
249 Cc: Thomas Gleixner <tglx@linutronix.de>
250 Cc: stable@kernel.org # reverted commits were marked for stable
251 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
252commit d966564fcdc19e13eb6ba1fbe6b8101070339c3d
253Author: Linus Torvalds <torvalds@linux-foundation.org>
254Date: Wed Feb 8 18:08:29 2017 -0800
255
256 Revert "x86/ioapic: Restore IO-APIC irq_chip retrigger callback"
257
258 This reverts commit 020eb3daaba2857b32c4cf4c82f503d6a00a67de.
259
260 Gabriel C reports that it causes his machine to not boot, and we haven't
261 tracked down the reason for it yet. Since the bug it fixes has been
262 around for a longish time, we're better off reverting the fix for now.
263
264 Gabriel says:
265 "It hangs early and freezes with a lot RCU warnings.
266
267 I bisected it down to :
268
269 > Ruslan Ruslichenko (1):
270 > x86/ioapic: Restore IO-APIC irq_chip retrigger callback
271
272 Reverting this one fixes the problem for me..
273
274 The box is a PRIMERGY TX200 S5 , 2 socket , 2 x E5520 CPU(s) installed"
275
276 and Ruslan and Thomas are currently stumped.
277
278 Reported-and-bisected-by: Gabriel C <nix.or.die@gmail.com>
279 Cc: Ruslan Ruslichenko <rruslich@cisco.com>
280 Cc: Thomas Gleixner <tglx@linutronix.de>
281 Cc: stable@kernel.org # for the backport of the original commit
282 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
283commit cdb98c2698b4af287925abcba4d77d92af82a0c3
284Author: Linus Torvalds <torvalds@linux-foundation.org>
285Date: Tue Dec 13 19:53:37 2016 -0800
286
287 Revert "nvme: add support for the Write Zeroes command"
288
289 This reverts commit 6d31e3ba232ea22458b2f36b6d3f2f9f11bf3fa4.
290
291 This causes bootup problems for me both on my laptop and my desktop.
292 What they have in common is that they have NVMe disks with dm-crypt, but
293 it's not the same controller, so it's not controller-specific.
294
295 Jens does not see it on his machine (also NVMe), so it's presumably
296 something that triggers just on bootup. Possibly related to dm-crypt
297 and the fact that I mark my luks volume with "allow-discards" in
298 /etc/crypttab.
299
300 It's 100% repeatable for me, which made it fairly straightforward to
301 bisect the problem to this commit. Small mercies.
302
303 So we don't know what the reason is yet, but the revert is needed to get
304 things going again.
305
306 Acked-by: Jens Axboe <axboe@fb.com>
307 Cc: Chaitanya Kulkarni <chaitanya.kulkarni@hgst.com>
308 Cc: Christoph Hellwig <hch@lst.de>
309 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
310commit 2b41226b39b654a5e20bce5a7332f307fdb9156b
311Author: Linus Torvalds <torvalds@linux-foundation.org>
312Date: Fri Dec 9 10:41:42 2016 -0800
313
314 Revert "radix tree test suite: fix compilation"
315
316 This reverts commit 53855d10f4567a0577360b6448d52a863929775b.
317
318 It shouldn't have come in yet - it depends on the changes in linux-next
319 that will come in during the next merge window. As Matthew Wilcox says,
320 the test suite is broken with the current state without the revert.
321
322 Requested-by: Matthew Wilcox <mawilcox@microsoft.com>
323 Cc: Andrew Morton <akpm@linux-foundation.org>
324 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
325commit f27c2f69cc8edc03ea8086f974811b9b45b2f3a5
326Author: Linus Torvalds <torvalds@linux-foundation.org>
327Date: Wed Dec 7 08:39:00 2016 -0800
328
329 Revert "default exported asm symbols to zero"
330
331 This reverts commit 8ab2ae655bfe384335c5b6b0d6041e0ddce26b00.
332
333 I loved that commit because of how it explained what the problem with
334 newer versions of binutils were, but the actual patch itself turns out
335 to not work very well.
336
337 It has two problems:
338
339 - a zero CRC value isn't actually right. It happens to work for the
340 case where both sides of the equation fail at giving the symbol a
341 crc, but there are cases where the users of the exported symbol get
342 the right crc (due to seeing the C declarations), but the actual
343 exporting itself does not (due to the whole weak asm symbol issue).
344
345 So then the module load fails after all - we did have a crc for the
346 symbol, but we couldn't match it with the loaded module.
347
348 - it seems that the alpha assembler has special semantics for the
349 '.set' directive, and on alpha it doesn't actually set the value of
350 the specified symbol at all, it is instead used to set various
351 assembly modes (eg ".set noat" and ".set noreorder").
352
353 So using ".set" to set the symbol value would just cause build
354 failures on alpha.
355
356 I'm sure we'll find some other workaround for these issues (hopefully
357 that involves getting rid of modversions entirely some day, but people
358 are also talking about just using smarter tools). But for now we'll
359 just fall back on commit faaae2a58143 ("Re-enable CONFIG_MODVERSIONS in
360 a slightly weaker form") that just let's a missing crc through.
361
362 Reported-by: Jan Stancek <jstancek@redhat.com>
363 Reported-by: Philip Müller <philm@manjaro.org>
364 Reported-by: Guenter Roeck <linux@roeck-us.net>
365 Cc: Arnd Bergmann <arnd@arndb.de>
366 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
367commit f5c9f9c72395c3291c2e35c905dedae2b98475a4
368Author: Linus Torvalds <torvalds@linux-foundation.org>
369Date: Mon Nov 14 09:31:52 2016 -0800
370
371 Revert "printk: make reading the kernel log flush pending lines"
372
373 This reverts commit bfd8d3f23b51018388be0411ccbc2d56277fe294.
374
375 It turns out that this flushes things much too aggressiverly, and causes
376 lines to break up when the system logger races with new continuation
377 lines being printed.
378
379 There's a pending patch to make printk() flushing much more
380 straightforward, but it's too invasive for 4.9, so in the meantime let's
381 just not make the system message logging flush continuation lines.
382 They'll be flushed by the final newline anyway.
383
384 Suggested-by: Petr Mladek <pmladek@suse.com>
385 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
386commit 38f7bd94a97b542de86a2be9229289717e33a7a4
387Author: Linus Torvalds <torvalds@linux-foundation.org>
388Date: Thu Sep 1 14:56:49 2016 -0700
389
390 Revert "af_unix: Fix splice-bind deadlock"
391
392 This reverts commit c845acb324aa85a39650a14e7696982ceea75dc1.
393
394 It turns out that it just replaces one deadlock with another one: we can
395 still get the wrong lock ordering with the readlock due to overlayfs
396 calling back into the filesystem layer and still taking the vfs locks
397 after the readlock.
398
399 The proper solution ends up being to just split the readlock into two
400 pieces: the bind lock (taken *outside* the vfs locks) and the IO lock
401 (taken *inside* the filesystem locks). The two locks are independent
402 anyway.
403
404 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
405 Reviewed-by: Shmulik Ladkani <shmulik.ladkani@gmail.com>
406 Signed-off-by: David S. Miller <davem@davemloft.net>
407commit a0cba2179ea4c1820fce2ee046b6ed90ecc56196
408Author: Linus Torvalds <torvalds@linux-foundation.org>
409Date: Tue Aug 9 10:48:18 2016 -0700
410
411 Revert "printk: create pr_<level> functions"
412
413 This reverts commit 874f9c7da9a4acbc1b9e12ca722579fb50e4d142.
414
415 Geert Uytterhoeven reports:
416 "This change seems to have an (unintendent?) side-effect.
417
418 Before, pr_*() calls without a trailing newline characters would be
419 printed with a newline character appended, both on the console and in
420 the output of the dmesg command.
421
422 After this commit, no new line character is appended, and the output
423 of the next pr_*() call of the same type may be appended, like in:
424
425 - Truncating RAM at 0x0000000040000000-0x00000000c0000000 to -0x0000000070000000
426 - Ignoring RAM at 0x0000000200000000-0x0000000240000000 (!CONFIG_HIGHMEM)
427 + Truncating RAM at 0x0000000040000000-0x00000000c0000000 to -0x0000000070000000Ignoring RAM at 0x0000000200000000-0x0000000240000000 (!CONFIG_HIGHMEM)"
428
429 Joe Perches says:
430 "No, that is not intentional.
431
432 The newline handling code inside vprintk_emit is a bit involved and
433 for now I suggest a revert until this has all the same behavior as
434 earlier"
435
436 Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
437 Requested-by: Joe Perches <joe@perches.com>
438 Cc: Andrew Morton <akpm@linux-foundation.org>
439 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
440commit 96b585267f552d4b6a28ea8bd75e5ed03deb6e71
441Author: Linus Torvalds <torvalds@linux-foundation.org>
442Date: Wed Aug 3 22:20:22 2016 -0400
443
444 Revert "ACPI / hotplug / PCI: Runtime resume bridge before rescan"
445
446 This reverts commit 16468c783cb4cf72475dcda23fabecb4a4bb0e17.
447
448 Bisection showed that it was the root cause for a resume hang on a
449 bog-standard all-Intel laptop (Sony Vaio Pro 11), and reverting fixes
450 the hang.
451
452 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
453commit 20d00ee829428ea6aab77fa3acca048a6f57d3bc
454Author: Linus Torvalds <torvalds@linux-foundation.org>
455Date: Fri Jul 29 12:17:52 2016 -0700
456
457 Revert "vfs: add lookup_hash() helper"
458
459 This reverts commit 3c9fe8cdff1b889a059a30d22f130372f2b3885f.
460
461 As Miklos points out in commit c1b2cc1a765a, the "lookup_hash()" helper
462 is now unused, and in fact, with the hash salting changes, since the
463 hash of a dentry name now depends on the directory dentry it is in, the
464 helper function isn't even really likely to be useful.
465
466 So rather than keep it around in case somebody else might end up finding
467 a use for it, let's just remove the helper and not trick people into
468 thinking it might be a useful thing.
469
470 For example, I had obviously completely missed how the helper didn't
471 follow the normal dentry hashing patterns, and how the hash salting
472 patch broke overlayfs. Things would quietly build and look sane, but
473 not work.
474
475 Suggested-by: Miklos Szeredi <mszeredi@redhat.com>
476 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
477commit 4046d6e81f33b7ef50d6668b78076d54c5e066b6
478Author: Linus Torvalds <torvalds@linux-foundation.org>
479Date: Thu Apr 14 11:18:57 2016 -0700
480
481 Revert "x86: remove the kernel code/data/bss resources from /proc/iomem"
482
483 This reverts commit c4004b02f8e5b9ce357a0bb1641756cc86962664.
484
485 Sadly, my hope that nobody would actually use the special kernel entries
486 in /proc/iomem were dashed by kexec. Which reads /proc/iomem explicitly
487 to find the kernel base address. Nasty.
488
489 Anyway, that means we can't do the sane and simple thing and just remove
490 the entries, and we'll instead have to mask them out based on permissions.
491
492 Reported-by: Zhengyu Zhang <zhezhang@redhat.com>
493 Reported-by: Dave Young <dyoung@redhat.com>
494 Reported-by: Freeman Zhang <freeman.zhang1992@gmail.com>
495 Reported-by: Emrah Demir <ed@abdsec.com>
496 Reported-by: Baoquan He <bhe@redhat.com>
497 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
498commit 9f2394c9be47a754bae9e4b6d382bdd4d77d0a11
499Author: Linus Torvalds <torvalds@linux-foundation.org>
500Date: Sun Apr 10 16:52:24 2016 -0700
501
502 Revert "ext4: allow readdir()'s of large empty directories to be interrupted"
503
504 This reverts commit 1028b55bafb7611dda1d8fed2aeca16a436b7dff.
505
506 It's broken: it makes ext4 return an error at an invalid point, causing
507 the readdir wrappers to write the the position of the last successful
508 directory entry into the position field, which means that the next
509 readdir will now return that last successful entry _again_.
510
511 You can only return fatal errors (that terminate the readdir directory
512 walk) from within the filesystem readdir functions, the "normal" errors
513 (that happen when the readdir buffer fills up, for example) happen in
514 the iterorator where we know the position of the actual failing entry.
515
516 I do have a very different patch that does the "signal_pending()"
517 handling inside the iterator function where it is allowable, but while
518 that one passes all the sanity checks, I screwed up something like four
519 times while emailing it out, so I'm not going to commit it today.
520
521 So my track record is not good enough, and the stars will have to align
522 better before that one gets committed. And it would be good to get some
523 review too, of course, since celestial alignments are always an iffy
524 debugging model.
525
526 IOW, let's just revert the commit that caused the problem for now.
527
528 Reported-by: Greg Thelen <gthelen@google.com>
529 Cc: Theodore Ts'o <tytso@mit.edu>
530 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
531commit 1701f680407c6cc05121566cef5293d33881fe71
532Author: Linus Torvalds <torvalds@linux-foundation.org>
533Date: Fri Mar 25 09:02:13 2016 -0700
534
535 Revert "ppdev: use new parport device model"
536
537 This reverts commit e7223f18603374d235d8bb0398532323e5f318b9.
538
539 It causes problems when a ppdev tries to register before the parport
540 driver has been registered with the device model. That will trigger the
541
542 BUG_ON(!drv->bus->p);
543
544 at drivers/base/driver.c:153. The call chain is
545
546 kernel_init ->
547 kernel_init_freeable ->
548 do_one_initcall ->
549 ppdev_init ->
550 __parport_register_driver ->
551 driver_register *BOOM*
552
553 Reported-by: kernel test robot <fengguang.wu@intel.com>
554 Reported-by: Ross Zwisler <zwisler@gmail.com>
555 Reported-by: Petr Mladek <pmladek@suse.com>
556 Cc: Sudip Mukherjee <sudipm.mukherjee@gmail.com>
557 Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
558 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
559commit 256faedcfd646161477d47a1a78c32a562d2e845
560Author: Linus Torvalds <torvalds@linux-foundation.org>
561Date: Mon Mar 7 13:15:09 2016 -0800
562
563 Revert "drm/radeon: call hpd_irq_event on resume"
564
565 This reverts commit dbb17a21c131eca94eb31136eee9a7fe5aff00d9.
566
567 It turns out that commit can cause problems for systems with multiple
568 GPUs, and causes X to hang on at least a HP Pavilion dv7 with hybrid
569 graphics.
570
571 This got noticed originally in 4.4.4, where this patch had already
572 gotten back-ported, but 4.5-rc7 was verified to have the same problem.
573
574 Alexander Deucher says:
575 "It looks like you have a muxed system so I suspect what's happening is
576 that one of the display is being reported as connected for both the
577 IGP and the dGPU and then the desktop environment gets confused or
578 there some sort problem in the detect functions since the mux is not
579 switched to the dGPU. I don't see an easy fix unless Dave has any
580 ideas. I'd say just revert for now"
581
582 Reported-by: Jörg-Volker Peetz <jvpeetz@web.de>
583 Acked-by: Alexander Deucher <Alexander.Deucher@amd.com>
584 Cc: Dave Airlie <airlied@gmail.com>
585 Cc: stable@kernel.org # wherever dbb17a21c131 got back-ported
586 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
587commit 00a3d660cbac05af34cca149cb80fb611e916935
588Author: Linus Torvalds <torvalds@linux-foundation.org>
589Date: Wed Oct 7 08:32:38 2015 +0100
590
591 Revert "fs: do not prefault sys_write() user buffer pages"
592
593 This reverts commit 998ef75ddb5709bbea0bf1506cd2717348a3c647.
594
595 The commit itself does not appear to be buggy per se, but it is exposing
596 a bug in ext4 (and Ted thinks ext3 too, but we solved that by getting
597 rid of it). It's too late in the release cycle to really worry about
598 this, even if Dave Hansen has a patch that may actually fix the
599 underlying ext4 problem. We can (and should) revisit this for the next
600 release.
601
602 The problem is that moving the prefaulting later now exposes a special
603 case with partially successful writes that isn't handled correctly. And
604 the prefaulting likely isn't normally even that much of a performance
605 issue - it looks like at least one reason Dave saw this in his
606 performance tests is that he also ran them on Skylake that now supports
607 the new SMAP code, which makes the normally very cheap user space
608 prefaulting noticeably more expensive.
609
610 Bisected-and-acked-by: Ted Ts'o <tytso@mit.edu>
611 Analyzed-and-acked-by: Dave Hansen <dave.hansen@linux.intel.com>
612 Cc: Andrew Morton <akpm@linux-foundation.org>
613 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
614commit 0ba13fd19d39b7cb672bcec052bc813389c079a4
615Author: Linus Torvalds <torvalds@linux-foundation.org>
616Date: Fri Sep 11 13:26:39 2015 -0700
617
618 Revert "writeback: plug writeback at a high level"
619
620 This reverts commit d353d7587d02116b9732d5c06615aed75a4d3a47.
621
622 Doing the block layer plug/unplug inside writeback_sb_inodes() is
623 broken, because that function is actually called with a spinlock held:
624 wb->list_lock, as pointed out by Chris Mason.
625
626 Chris suggested just dropping and re-taking the spinlock around the
627 blk_finish_plug() call (the plgging itself can happen under the
628 spinlock), and that would technically work, but is just disgusting.
629
630 We do something fairly similar - but not quite as disgusting because we
631 at least have a better reason for it - in writeback_single_inode(), so
632 it's not like the caller can depend on the lock being held over the
633 call, but in this case there just isn't any good reason for that
634 "release and re-take the lock" pattern.
635
636 [ In general, we should really strive to avoid the "release and retake"
637 pattern for locks, because in the general case it can easily cause
638 subtle bugs when the caller caches any state around the call that
639 might be invalidated by dropping the lock even just temporarily. ]
640
641 But in this case, the plugging should be easy to just move up to the
642 callers before the spinlock is taken, which should even improve the
643 effectiveness of the plug. So there is really no good reason to play
644 games with locking here.
645
646 I'll send off a test-patch so that Dave Chinner can verify that that
647 plug movement works. In the meantime this just reverts the problematic
648 commit and adds a comment to the function so that we hopefully don't
649 make this mistake again.
650
651 Reported-by: Chris Mason <clm@fb.com>
652 Cc: Josef Bacik <jbacik@fb.com>
653 Cc: Dave Chinner <david@fromorbit.com>
654 Cc: Neil Brown <neilb@suse.de>
655 Cc: Jan Kara <jack@suse.cz>
656 Cc: Christoph Hellwig <hch@lst.de>
657 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
658commit ed596cde9425509ec6ce88e19f03e9b13b6f518b
659Author: Linus Torvalds <torvalds@linux-foundation.org>
660Date: Thu Aug 13 08:25:20 2015 -0700
661
662 Revert x86 sigcontext cleanups
663
664 This reverts commits 9a036b93a344 ("x86/signal/64: Remove 'fs' and 'gs'
665 from sigcontext") and c6f2062935c8 ("x86/signal/64: Fix SS handling for
666 signals delivered to 64-bit programs").
667
668 They were cleanups, but they break dosemu by changing the signal return
669 behavior (and removing 'fs' and 'gs' from the sigcontext struct - while
670 not actually changing any behavior - causes build problems).
671
672 Reported-and-tested-by: Stas Sergeev <stsp@list.ru>
673 Acked-by: Andy Lutomirski <luto@amacapital.net>
674 Cc: Ingo Molnar <mingo@kernel.org>
675 Cc: stable@vger.kernel.org
676 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
677commit d725e66c06ab440032f49ef17e960896d0ec6d49
678Author: Linus Torvalds <torvalds@linux-foundation.org>
679Date: Tue Jul 21 16:06:53 2015 -0700
680
681 Revert "fsnotify: fix oops in fsnotify_clear_marks_by_group_flags()"
682
683 This reverts commit a2673b6e040663bf16a552f8619e6bde9f4b9acf.
684
685 Kinglong Mee reports a memory leak with that patch, and Jan Kara confirms:
686
687 "Thanks for report! You are right that my patch introduces a race
688 between fsnotify kthread and fsnotify_destroy_group() which can result
689 in leaking inotify event on group destruction.
690
691 I haven't yet decided whether the right fix is not to queue events for
692 dying notification group (as that is pointless anyway) or whether we
693 should just fix the original problem differently... Whenever I look
694 at fsnotify code mark handling I get lost in the maze of locks, lists,
695 and subtle differences between how different notification systems
696 handle notification marks :( I'll think about it over night"
697
698 and after thinking about it, Jan says:
699
700 "OK, I have looked into the code some more and I found another
701 relatively simple way of fixing the original oops. It will be IMHO
702 better than trying to fixup this issue which has more potential for
703 breakage. I'll ask Linus to revert the fsnotify fix he already merged
704 and send a new fix"
705
706 Reported-by: Kinglong Mee <kinglongmee@gmail.com>
707 Requested-by: Jan Kara <jack@suse.cz>
708 Cc: Andrew Morton <akpm@linux-foundation.org>
709 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
710commit 01e2d0627a9a6edb24c37db45db5ecb31e9de808
711Author: Linus Torvalds <torvalds@linux-foundation.org>
712Date: Sun Jul 12 15:00:20 2015 -0700
713
714 Revert "drm/i915: Use crtc_state->active in primary check_plane func"
715
716 This reverts commit dec4f799d0a4c9edae20512fa60b0a36f3299ca2.
717
718 Jörg Otte reports a NULL pointder dereference due to this commit, as
719 'crtc_state' very much can be NULL:
720
721 crtc_state = state->base.state ?
722 intel_atomic_get_crtc_state(state->base.state, intel_crtc) : NULL;
723
724 So the change to test 'crtc_state->base.active' cannot possibly be
725 correct as-is.
726
727 There may be some other minimal fix (like just checking crtc_state for
728 NULL), but I'm just reverting it now for the rc2 release, and people
729 like Daniel Vetter who actually know this code will figure out what the
730 right solution is in the longer term.
731
732 Reported-and-bisected-by: Jörg Otte <jrg.otte@gmail.com>
733 Cc: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
734 Cc: Jani Nikula <jani.nikula@linux.intel.com>
735 Cc: Daniel Vetter <daniel.vetter@intel.com>
736 CC: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
737 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
738commit ebeaa8ddb3663b5c6cfc205605c35116381550c5
739Author: Linus Torvalds <torvalds@linux-foundation.org>
740Date: Sat Jun 27 09:41:50 2015 -0700
741
742 Revert "jbd2: speedup jbd2_journal_dirty_metadata()"
743
744 This reverts commit 2143c1965a761332ae417b22fd477b636e4f54ec.
745
746 This commit seems to be the cause of the following jbd2 assertion
747 failure:
748
749 ------------[ cut here ]------------
750 kernel BUG at fs/jbd2/transaction.c:1325!
751 invalid opcode: 0000 [#1] SMP
752 Modules linked in: bnep bluetooth fuse ip6t_rpfilter ip6t_REJECT nf_reject_ipv6 nf_conntrack_ipv6 ...
753 CPU: 7 PID: 5509 Comm: gcc Not tainted 4.1.0-10944-g2a298679b411 #1
754 Hardware name: /DH87RL, BIOS RLH8710H.86A.0327.2014.0924.1645 09/24/2014
755 task: ffff8803bf866040 ti: ffff880308528000 task.ti: ffff880308528000
756 RIP: jbd2_journal_dirty_metadata+0x237/0x290
757 Call Trace:
758 __ext4_handle_dirty_metadata+0x43/0x1f0
759 ext4_handle_dirty_dirent_node+0xde/0x160
760 ? jbd2_journal_get_write_access+0x36/0x50
761 ext4_delete_entry+0x112/0x160
762 ? __ext4_journal_start_sb+0x52/0xb0
763 ext4_unlink+0xfa/0x260
764 vfs_unlink+0xec/0x190
765 do_unlinkat+0x24a/0x270
766 SyS_unlink+0x11/0x20
767 entry_SYSCALL_64_fastpath+0x12/0x6a
768 ---[ end trace ae033ebde8d080b4 ]---
769
770 which is not easily reproducible (I've seen it just once, and then Ted
771 was able to reproduce it once). Revert it while Ted and Jan try to
772 figure out what is wrong.
773
774 Cc: Jan Kara <jack@suse.cz>
775 Acked-by: Theodore Ts'o <tytso@mit.edu>
776 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
777commit f3ca10dde49043fbbb055854278b26954b549b36
778Author: Linus Torvalds <torvalds@linux-foundation.org>
779Date: Wed Apr 22 09:44:36 2015 -0700
780
781 Revert "mm: avoid tail page refcounting on non-THP compound pages"
782
783 This reverts commit 8d63d99a5dfbdb997d12dd3c07b2070ca723db3b.
784
785 It causes in VM mapping refcount errors:
786
787 page:ffffea0010a15040 count:0 mapcount:1 mapping: (null) index:0x0
788 flags: 0x8000000000008014(referenced|dirty|tail)
789 page dumped because: VM_BUG_ON_PAGE(page_mapcount(page) != 0)
790 ------------[ cut here ]------------
791 kernel BUG at mm/swap.c:134!
792
793 as reported by Borislav Petkov
794
795 Reported-and-tested-by: Borislav Petkov <bp@alien8.de>
796 Cc: Kirill A. Shutemov <kirill@shutemov.name>
797 Cc: Hugh Dickins <hughd@google.com>
798 Cc: Andrea Arcangeli <aarcange@redhat.com>
799 Cc: Andrew Morton <akpm@linux-foundation.org>
800 Cc: linux-mm@kvack.org
801 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
802commit 8f443e2372ba23d51ee365974f54507acd6f69d1
803Author: Linus Torvalds <torvalds@linux-foundation.org>
804Date: Tue Apr 21 09:17:28 2015 -0700
805
806 Revert "ocfs2: incorrect check for debugfs returns"
807
808 This reverts commit e2ac55b6a8e337fac7cc59c6f452caac92ab5ee6.
809
810 Huang Ying reports that this causes a hang at boot with debugfs disabled.
811
812 It is true that the debugfs error checks are kind of confusing, and this
813 code certainly merits more cleanup and thinking about it, but there's
814 something wrong with the trivial "check not just for NULL, but for error
815 pointers too" patch.
816
817 Yes, with debugfs disabled, we will end up setting the o2hb_debug_dir
818 pointer variable to an error pointer (-ENODEV), and then continue as if
819 everything was fine. But since debugfs is disabled, all the _users_ of
820 that pointer end up being compiled away, so even though the pointer can
821 not be dereferenced, that's still fine.
822
823 So it's confusing and somewhat questionable, but the "more correct"
824 error checks end up causing more trouble than they fix.
825
826 Reported-by: Huang Ying <ying.huang@intel.com>
827 Acked-by: Andrew Morton <akpm@linux-foundation.org>
828 Acked-by: Chengyu Song <csong84@gatech.edu>
829 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
830commit 12522eeac88165d0db56a6a19ab607addf728995
831Author: Linus Torvalds <torvalds@linux-foundation.org>
832Date: Sat Apr 11 13:46:07 2015 -0700
833
834 Revert "dmaengine: Add a warning for drivers not using the generic slave caps retrieval"
835
836 This reverts commit ecc19d17868be9c9f8f00ed928791533c420f3e0.
837
838 It added a new warning to try to encourage driver writers to set the
839 device capabities properly, but drivers haven't been updated and in the
840 meantime it just generaters a scary message that users cannot actually
841 do anything about.
842
843 Warnings like these are appropriate if you actually expect to fix the
844 code that causes them. They are not appropriate for releases.
845
846 Requested-by: Peter Hurley <peter@hurleysoftware.com>
847 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
848commit 8329aa9fff3fca84009e6a444d8d160193643bac
849Author: Linus Torvalds <torvalds@linux-foundation.org>
850Date: Fri Feb 13 10:26:18 2015 -0800
851
852 Revert "x86/apic: Only disable CPU x2apic mode when necessary"
853
854 This reverts commit 5fcee53ce705d49c766f8a302c7e93bdfc33c124.
855
856 It causes the suspend to fail on at least the Chromebook Pixel, possibly
857 other platforms too.
858
859 Joerg Roedel points out that the logic should probably have been
860
861 if (max_physical_apicid > 255 ||
862 !(IS_ENABLED(CONFIG_HYPERVISOR_GUEST) &&
863 hypervisor_x2apic_available())) {
864
865 instead, but since the code is not in any fast-path, so we can just live
866 without that optimization and just revert to the original code.
867
868 Acked-by: Joerg Roedel <joro@8bytes.org>
869 Acked-by: Jiang Liu <jiang.liu@linux.intel.com>
870 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
871commit 7f4054836d811c650c51f9c93088f8ebd61b0020
872Author: Linus Torvalds <torvalds@linux-foundation.org>
873Date: Wed Dec 31 12:59:34 2014 -0800
874
875 Revert "Input: atmel_mxt_ts - use deep sleep mode when stopped"
876
877 This reverts commit 9d469d033d135d80742a4e39e6bbb4519dd5eee1.
878
879 It breaks the Chromebook Pixel touchpad (and touchscreen).
880
881 Reported-by: Dirk Hohndel <dirk@hohndel.org>
882 Bisected-by: Linus Torvalds <torvalds@linux-foundation.org>
883 Cc: Nick Dyer <nick.dyer@itdev.co.uk>
884 Cc: Benson Leung <bleung@chromium.org>
885 Cc: Yufeng Shen <miletus@chromium.org>
886 Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
887 Cc: stable@vger.kernel.org # v3.16+
888 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
889commit f3670394c29ff3730638762c1760fd2f624e6d7b
890Author: Linus Torvalds <torvalds@linux-foundation.org>
891Date: Mon Sep 22 23:05:49 2014 -0700
892
893 Revert "x86/efi: Fixup GOT in all boot code paths"
894
895 This reverts commit 9cb0e394234d244fe5a97e743ec9dd7ddff7e64b.
896
897 It causes my Sony Vaio Pro 11 to immediately reboot at startup.
898
899 Acked-by: Ingo Molnar <mingo@kernel.org>
900 Cc: Peter Anvin <hpa@zytor.com>
901 Cc: Maarten Lankhorst <maarten.lankhorst@canonical.com>
902 Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
903 Cc: Matt Fleming <matt.fleming@intel.com>
904 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
905commit 155134fef2b6c7426c3f25ffe84fb3043167c860
906Author: Linus Torvalds <torvalds@linux-foundation.org>
907Date: Sun Aug 10 21:24:59 2014 -0700
908
909 Revert "proc: Point /proc/{mounts,net} at /proc/thread-self/{mounts,net} instead of /proc/self/{mounts,net}"
910
911 This reverts commits 344470cac42e and e81324407269.
912
913 It turns out that the exact path in the symlink matters, if for somewhat
914 unfortunate reasons: some apparmor configurations don't allow dhclient
915 access to the per-thread /proc files. As reported by Jörg Otte:
916
917 audit: type=1400 audit(1407684227.003:28): apparmor="DENIED"
918 operation="open" profile="/sbin/dhclient"
919 name="/proc/1540/task/1540/net/dev" pid=1540 comm="dhclient"
920 requested_mask="r" denied_mask="r" fsuid=0 ouid=0
921
922 so we had better revert this for now. We might be able to work around
923 this in practice by only using the per-thread symlinks if the thread
924 isn't the thread group leader, and if the namespaces differ between
925 threads (which basically never happens).
926
927 We'll see. In the meantime, the revert was made to be intentionally easy.
928
929 Reported-by: Jörg Otte <jrg.otte@gmail.com>
930 Acked-by: Eric W. Biederman <ebiederm@xmission.com>
931 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
932commit 27d438c56009e5ae632de36fe70985d1aab5e344
933Author: Linus Torvalds <torvalds@linux-foundation.org>
934Date: Fri Aug 8 11:48:26 2014 -1000
935
936 Revert "drm/i915: Enable PSR by default."
937
938 This reverts commit b6d547791fd3ef4ccc89ad2556ab01045640aef7.
939
940 The panel self refresh clearly isn't stable yet, and causes my laptop
941 (Haswell ULT in a Sony Vaio Pro) to have the screen lock up. Maybe it
942 doesn't ever get out of self-refresh, or maybe there are gremlins in the
943 machine that get unhappy. Regardless, it's broken, and it gets
944 reverted.
945
946 Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
947 Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
948 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
949commit 1d8fcba1de632d7a43349788ad534c5a32c5a44c
950Author: Linus Torvalds <torvalds@linux-foundation.org>
951Date: Wed Jul 30 08:56:23 2014 -0700
952
953 Revert "cdc_subset: deal with a device that needs reset for timeout"
954
955 This reverts commit 20fbe3ae990fd54fc7d1f889d61958bc8b38f254.
956
957 As reported by Stephen Rothwell, it causes compile failures in certain
958 configurations:
959
960 drivers/net/usb/cdc_subset.c:360:15: error: 'dummy_prereset' undeclared here (not in a function)
961 .pre_reset = dummy_prereset,
962 ^
963 drivers/net/usb/cdc_subset.c:361:16: error: 'dummy_postreset' undeclared here (not in a function)
964 .post_reset = dummy_postreset,
965 ^
966
967 Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
968 Acked-by: David Miller <davem@davemloft.net>
969 Cc: Oliver Neukum <oneukum@suse.de>
970 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
971commit bb077d600689dbf9305758efed1e16775db1c84c
972Author: Linus Torvalds <torvalds@linux-foundation.org>
973Date: Sun Jun 8 10:09:49 2014 -0700
974
975 Revert "x86/smpboot: Initialize secondary CPU only if master CPU will wait for it"
976
977 This reverts commit 3e1a878b7ccdb31da6d9d2b855c72ad87afeba3f.
978
979 It came in very late, and already has one reported failure: Sitsofe
980 reports that the current tree fails to boot on his EeePC, and bisected
981 it down to this. Rather than waste time trying to figure out what's
982 wrong, just revert it.
983
984 Reported-by: Sitsofe Wheeler <sitsofe@gmail.com>
985 Cc: Igor Mammedov <imammedo@redhat.com>
986 Cc: Toshi Kani <toshi.kani@hp.com>
987 Cc: Thomas Gleixner <tglx@linutronix.de>
988 Acked-by: Ingo Molnar <mingo@kernel.org>
989 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
990commit 29b1deb2a48a9dd02b93597aa4c055a24c0e989f
991Author: Linus Torvalds <torvalds@linux-foundation.org>
992Date: Sun Dec 15 11:17:45 2013 -0800
993
994 Revert "selinux: consider filesystem subtype in policies"
995
996 This reverts commit 102aefdda4d8275ce7d7100bc16c88c74272b260.
997
998 Tom London reports that it causes sync() to hang on Fedora rawhide:
999
1000 https://bugzilla.redhat.com/show_bug.cgi?id=1033965
1001
1002 and Josh Boyer bisected it down to this commit. Reverting the commit in
1003 the rawhide kernel fixes the problem.
1004
1005 Eric Paris root-caused it to incorrect subtype matching in that commit
1006 breaking fuse, and has a tentative patch, but by now we're better off
1007 retrying this in 3.14 rather than playing with it any more.
1008
1009 Reported-by: Tom London <selinux@gmail.com>
1010 Bisected-by: Josh Boyer <jwboyer@fedoraproject.org>
1011 Acked-by: Eric Paris <eparis@redhat.com>
1012 Cc: James Morris <jmorris@namei.org>
1013 Cc: Anand Avati <avati@redhat.com>
1014 Cc: Paul Moore <paul@paul-moore.com>
1015 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1016commit 4c1cc40a2d49500d84038ff751bc6cd183e729b5
1017Author: Linus Torvalds <torvalds@linux-foundation.org>
1018Date: Sat Nov 23 16:38:17 2013 -0800
1019
1020 Revert "KEYS: verify a certificate is signed by a 'trusted' key"
1021
1022 This reverts commit 09fbc47373826d67531380662b516de2da120545, which
1023 caused the following build errors:
1024
1025 crypto/asymmetric_keys/x509_public_key.c: In function ‘x509_key_preparse’:
1026 crypto/asymmetric_keys/x509_public_key.c:237:35: error: ‘system_trusted_keyring’ undeclared (first use in this function)
1027 ret = x509_validate_trust(cert, system_trusted_keyring);
1028 ^
1029 crypto/asymmetric_keys/x509_public_key.c:237:35: note: each undeclared identifier is reported only once for each function it appears in
1030
1031 reported by Jim Davis. Mimi says:
1032
1033 "I made the classic mistake of requesting this patch to be upstreamed
1034 at the last second, rather than waiting until the next open window.
1035
1036 At this point, the best course would probably be to revert the two
1037 commits and fix them for the next open window"
1038
1039 Reported-by: Jim Davis <jim.epost@gmail.com>
1040 Acked-by: Mimi Zohar <zohar@linux.vnet.ibm.com>
1041 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1042commit 34ef7bd3823bf4401bf8f1f855e1bc77b82b1a43
1043Author: Linus Torvalds <torvalds@linux-foundation.org>
1044Date: Sat Nov 23 16:36:35 2013 -0800
1045
1046 Revert "ima: define '_ima' as a builtin 'trusted' keyring"
1047
1048 This reverts commit 217091dd7a7a1bdac027ddb7c5a25f6ac0b8e241, which
1049 caused the following build error:
1050
1051 security/integrity/digsig.c:70:5: error: redefinition of ‘integrity_init_keyring’
1052 security/integrity/integrity.h:149:12: note: previous definition of ‘integrity_init_keyring’ w
1053 security/integrity/integrity.h:149:12: warning: ‘integrity_init_keyring’ defined but not used
1054
1055 reported by Krzysztof Kolasa. Mimi says:
1056
1057 "I made the classic mistake of requesting this patch to be upstreamed
1058 at the last second, rather than waiting until the next open window.
1059
1060 At this point, the best course would probably be to revert the two
1061 commits and fix them for the next open window"
1062
1063 Reported-by: Krzysztof Kolasa <kkolasa@winsoft.pl>
1064 Acked-by: Mimi Zohar <zohar@linux.vnet.ibm.com>
1065 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1066commit 8b2e9b712f6139df9c754af0d67fecc4bbc88545
1067Author: Linus Torvalds <torvalds@linux-foundation.org>
1068Date: Wed Nov 20 14:41:47 2013 -0800
1069
1070 Revert "mm: create a separate slab for page->ptl allocation"
1071
1072 This reverts commit ea1e7ed33708c7a760419ff9ded0a6cb90586a50.
1073
1074 Al points out that while the commit *does* actually create a separate
1075 slab for the page->ptl allocation, that slab is never actually used, and
1076 the code continues to use kmalloc/kfree.
1077
1078 Damien Wyart points out that the original patch did have the conversion
1079 to use kmem_cache_alloc/free, so it got lost somewhere on its way to me.
1080
1081 Revert the half-arsed attempt that didn't do anything. If we really do
1082 want the special slab (remember: this is all relevant just for debug
1083 builds, so it's not necessarily all that critical) we might as well redo
1084 the patch fully.
1085
1086 Reported-by: Al Viro <viro@zeniv.linux.org.uk>
1087 Acked-by: Andrew Morton <akpm@linux-foundation.org>
1088 Cc: Kirill A Shutemov <kirill.shutemov@linux.intel.com>
1089 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1090commit a1212d278c05ca0a38f5cbd7ae90ac2e367228a8
1091Author: Linus Torvalds <torvalds@linux-foundation.org>
1092Date: Thu Nov 7 20:47:28 2013 +0900
1093
1094 Revert "sysfs: drop kobj_ns_type handling"
1095
1096 This reverts commit cb26a311578e67769e92a39a0a63476533cb7e12.
1097
1098 It mysteriously causes NetworkManager to not find the wireless device
1099 for me. As far as I can tell, Tejun *meant* for this commit to not make
1100 any semantic changes, but there clearly are some. So revert it, taking
1101 into account some of the calling convention changes that happened in
1102 this area in subsequent commits.
1103
1104 Cc: Tejun Heo <tj@kernel.org>
1105 Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
1106 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1107commit b04c99e3b845892d754ee8052d6324c39c4040de
1108Author: Linus Torvalds <torvalds@linux-foundation.org>
1109Date: Sat Sep 7 09:48:41 2013 -0700
1110
1111 Revert "Input: introduce BTN/ABS bits for drums and guitars"
1112
1113 This reverts commits 61e00655e9cb, 73f8645db191 and 8e22ecb603c8:
1114 "Input: introduce BTN/ABS bits for drums and guitars"
1115 "HID: wiimote: add support for Guitar-Hero drums"
1116 "HID: wiimote: add support for Guitar-Hero guitars"
1117
1118 The extra new ABS_xx values resulted in ABS_MAX no longer being a
1119 power-of-two, which broke the comparison logic. It also caused the
1120 ioctl numbers to overflow into the next byte, causing problems for that.
1121
1122 We'll try again for 3.13.
1123
1124 Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de>
1125 Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
1126 Acked-by: David Herrmann <dh.herrmann@gmail.com>
1127 Acked-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
1128 Cc: Benjamin Tissoires <benjamin.tissoires@gmail.com>
1129 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1130commit f0cc6ffb8ce8961db587e5072168cac0cbc25f05
1131Author: Linus Torvalds <torvalds@linux-foundation.org>
1132Date: Wed Aug 28 09:18:05 2013 -0700
1133
1134 Revert "fs: Allow unprivileged linkat(..., AT_EMPTY_PATH) aka flink"
1135
1136 This reverts commit bb2314b47996491bbc5add73633905c3120b6268.
1137
1138 It wasn't necessarily wrong per se, but we're still busily discussing
1139 the exact details of this all, so I'm going to revert it for now.
1140
1141 It's true that you can already do flink() through /proc and that flink()
1142 isn't new. But as Brad Spengler points out, some secure environments do
1143 not mount proc, and flink adds a new interface that can avoid path
1144 lookup of the source for those kinds of environments.
1145
1146 We may re-do this (and even mark it for stable backporting back in 3.11
1147 and possibly earlier) once the whole discussion about the interface is done.
1148
1149 Cc: Andy Lutomirski <luto@amacapital.net>
1150 Cc: Al Viro <viro@zeniv.linux.org.uk>
1151 Cc: Oleg Nesterov <oleg@redhat.com>
1152 Cc: Brad Spengler <spender@grsecurity.net>
1153 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1154commit 5ea80f76a56605a190a7ea16846c82aa63dbd0aa
1155Author: Linus Torvalds <torvalds@linux-foundation.org>
1156Date: Thu Aug 22 09:13:06 2013 -0700
1157
1158 Revert "x86 get_unmapped_area(): use proper mmap base for bottom-up direction"
1159
1160 This reverts commit df54d6fa54275ce59660453e29d1228c2b45a826.
1161
1162 The commit isn't necessarily wrong, but because it recalculates the
1163 random mmap_base every time, it seems to confuse user memory allocators
1164 that expect contiguous mmap allocations even when the mmap address isn't
1165 specified.
1166
1167 In particular, the MATLAB Java runtime seems to be unhappy. See
1168
1169 https://bugzilla.kernel.org/show_bug.cgi?id=60774
1170
1171 So we'll want to apply the random offset only once, and Radu has a patch
1172 for that. Revert this older commit in order to apply the other one.
1173
1174 Reported-by: Jeff Shorey <shoreyjeff@gmail.com>
1175 Cc: Radu Caragea <sinaelgl@gmail.com>
1176 Cc: Andrew Morton <akpm@linux-foundation.org>
1177 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1178commit 370905069ce6515f38d5de0a5b1c899cbe58fe22
1179Author: Linus Torvalds <torvalds@linux-foundation.org>
1180Date: Thu Aug 8 09:06:37 2013 -0700
1181
1182 Revert "slub: do not put a slab to cpu partial list when cpu_partial is 0"
1183
1184 This reverts commit 318df36e57c0ca9f2146660d41ff28e8650af423.
1185
1186 This commit caused Steven Rostedt's hackbench runs to run out of memory
1187 due to a leak. As noted by Joonsoo Kim, it is buggy in the following
1188 scenario:
1189
1190 "I guess, you may set 0 to all kmem caches's cpu_partial via sysfs,
1191 doesn't it?
1192
1193 In this case, memory leak is possible in following case. Code flow of
1194 possible leak is follwing case.
1195
1196 * in __slab_free()
1197 1. (!new.inuse || !prior) && !was_frozen
1198 2. !kmem_cache_debug && !prior
1199 3. new.frozen = 1
1200 4. after cmpxchg_double_slab, run the (!n) case with new.frozen=1
1201 5. with this patch, put_cpu_partial() doesn't do anything,
1202 because this cache's cpu_partial is 0
1203 6. return
1204
1205 In step 5, leak occur"
1206
1207 And Steven does indeed have cpu_partial set to 0 due to RT testing.
1208
1209 Joonsoo is cooking up a patch, but everybody agrees that reverting this
1210 for now is the right thing to do.
1211
1212 Reported-and-bisected-by: Steven Rostedt <rostedt@goodmis.org>
1213 Acked-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
1214 Acked-by: Pekka Enberg <penberg@kernel.org>
1215 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1216commit 6d128e1e72bf082542e85f72e6b7ddd704193588
1217Author: Linus Torvalds <torvalds@linux-foundation.org>
1218Date: Wed Jul 10 19:02:51 2013 -0700
1219
1220 Revert "Makefile: Fix install error with make -j option"
1221
1222 This reverts commit d2aae8477cd00325bb7c7c7e95be488088900c48.
1223
1224 It is completely and utterly broken. Module install should not build
1225 any files, and adding broken dependencies to "help" it build files is
1226 complete and utter sh*t.
1227
1228 The kernel should not be built by root, and "make install" and "make
1229 module_install" (that for obvious reasons need to be run as root)
1230 absolutely must not build any files. They should only ever copy the
1231 already-built files over.
1232
1233 So having dependencies for the install targets is wrong, wrong, wrong.
1234
1235 If you try to install a kernel without building it first, you *should*
1236 get errors. The build system shouldn't try to help root build the files.
1237
1238 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1239commit 0a82a8d132b26d438eb90b3ab35a7016e7227a1d
1240Author: Linus Torvalds <torvalds@linux-foundation.org>
1241Date: Thu Apr 18 09:00:26 2013 -0700
1242
1243 Revert "block: add missing block_bio_complete() tracepoint"
1244
1245 This reverts commit 3a366e614d0837d9fc23f78cdb1a1186ebc3387f.
1246
1247 Wanlong Gao reports that it causes a kernel panic on his machine several
1248 minutes after boot. Reverting it removes the panic.
1249
1250 Jens says:
1251 "It's not quite clear why that is yet, so I think we should just revert
1252 the commit for 3.9 final (which I'm assuming is pretty close).
1253
1254 The wifi is crap at the LSF hotel, so sending this email instead of
1255 queueing up a revert and pull request."
1256
1257 Reported-by: Wanlong Gao <gaowanlong@cn.fujitsu.com>
1258 Requested-by: Jens Axboe <axboe@kernel.dk>
1259 Cc: Tejun Heo <tj@kernel.org>
1260 Cc: Steven Rostedt <rostedt@goodmis.org>
1261 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1262commit 9360b53661a2c7754517b2925580055bacc8ec38
1263Author: Linus Torvalds <torvalds@linux-foundation.org>
1264Date: Mon Dec 17 11:29:09 2012 -0800
1265
1266 Revert "bdi: add a user-tunable cpu_list for the bdi flusher threads"
1267
1268 This reverts commit 8fa72d234da9b6b473bbb1f74d533663e4996e6b.
1269
1270 People disagree about how this should be done, so let's revert this for
1271 now so that nobody starts using the new tuning interface. Tejun is
1272 thinking about a more generic interface for thread pool affinity.
1273
1274 Requested-by: Tejun Heo <tj@kernel.org>
1275 Acked-by: Jeff Moyer <jmoyer@redhat.com>
1276 Acked-by: Jens Axboe <axboe@kernel.dk>
1277 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1278commit 11520e5e7c1855fc3bf202bb3be35a39d9efa034
1279Author: Linus Torvalds <torvalds@linux-foundation.org>
1280Date: Sat Dec 15 15:15:24 2012 -0800
1281
1282 Revert "x86-64/efi: Use EFI to deal with platform wall clock (again)"
1283
1284 This reverts commit bd52276fa1d4 ("x86-64/efi: Use EFI to deal with
1285 platform wall clock (again)"), and the two supporting commits:
1286
1287 da5a108d05b4: "x86/kernel: remove tboot 1:1 page table creation code"
1288
1289 185034e72d59: "x86, efi: 1:1 pagetable mapping for virtual EFI calls")
1290
1291 as they all depend semantically on commit 53b87cf088e2 ("x86, mm:
1292 Include the entire kernel memory map in trampoline_pgd") that got
1293 reverted earlier due to the problems it caused.
1294
1295 This was pointed out by Yinghai Lu, and verified by me on my Macbook Air
1296 that uses EFI.
1297
1298 Pointed-out-by: Yinghai Lu <yinghai@kernel.org>
1299 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1300commit be354f40812314dee2b1e3aa272528c056bb827d
1301Author: Linus Torvalds <torvalds@linux-foundation.org>
1302Date: Sat Dec 15 12:29:54 2012 -0800
1303
1304 Revert "x86, mm: Include the entire kernel memory map in trampoline_pgd"
1305
1306 This reverts commit 53b87cf088e2ea68d7c59619d0214cc15bb76133.
1307
1308 It causes odd bootup problems on x86-64. Markus Trippelsdorf gets a
1309 repeatable oops, and I see a non-repeatable oops (or constant stream of
1310 messages that scroll off too quickly to read) that seems to go away with
1311 this commit reverted.
1312
1313 So we don't know exactly what is wrong with the commit, but it's
1314 definitely problematic, and worth reverting sooner rather than later.
1315
1316 Bisected-by: Markus Trippelsdorf <markus@trippelsdorf.de>
1317 Cc: H Peter Anvin <hpa@zytor.com>
1318 Cc: Jan Beulich <jbeulich@suse.com>
1319 Cc: Matt Fleming <matt.fleming@intel.com>
1320 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1321commit 17bc14b767cf0692420c43dbe5310ae98a5a7836
1322Author: Linus Torvalds <torvalds@linux-foundation.org>
1323Date: Fri Dec 14 07:20:43 2012 -0800
1324
1325 Revert "sched: Update_cfs_shares at period edge"
1326
1327 This reverts commit f269ae0469fc882332bdfb5db15d3c1315fe2a10.
1328
1329 It turns out it causes a very noticeable interactivity regression with
1330 CONFIG_SCHED_AUTOGROUP (test-case: "make -j32" of the kernel in a
1331 terminal window, while scrolling in a browser - the autogrouping means
1332 that the two end up in separate cgroups, and the browser should be
1333 smooth as silk despite the high load).
1334
1335 Says Paul Turner:
1336 "It seems that the update-throttling on the wake-side is reducing the
1337 interactive tasks' ability to preempt. While I suspect the right
1338 longer term answer here is force these updates only in the
1339 cross-cgroup case; this is less trivial. For this release I believe
1340 the right answer is either going to be a revert or restore the updates
1341 on the enqueue-side."
1342
1343 Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
1344 Bisected-by: Mike Galbraith <efault@gmx.de>
1345 Acked-by: Paul Turner <pjt@google.com>
1346 Acked-by: Ingo Molnar <mingo@kernel.org>
1347 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1348commit caf491916b1c1e939a2c7575efb7a77f11fc9bdf
1349Author: Linus Torvalds <torvalds@linux-foundation.org>
1350Date: Mon Dec 10 10:51:16 2012 -0800
1351
1352 Revert "revert "Revert "mm: remove __GFP_NO_KSWAPD""" and associated damage
1353
1354 This reverts commits a50915394f1fc02c2861d3b7ce7014788aa5066e and
1355 d7c3b937bdf45f0b844400b7bf6fd3ed50bac604.
1356
1357 This is a revert of a revert of a revert. In addition, it reverts the
1358 even older i915 change to stop using the __GFP_NO_KSWAPD flag due to the
1359 original commits in linux-next.
1360
1361 It turns out that the original patch really was bogus, and that the
1362 original revert was the correct thing to do after all. We thought we
1363 had fixed the problem, and then reverted the revert, but the problem
1364 really is fundamental: waking up kswapd simply isn't the right thing to
1365 do, and direct reclaim sometimes simply _is_ the right thing to do.
1366
1367 When certain allocations fail, we simply should try some direct reclaim,
1368 and if that fails, fail the allocation. That's the right thing to do
1369 for THP allocations, which can easily fail, and the GPU allocations want
1370 to do that too.
1371
1372 So starting kswapd is sometimes simply wrong, and removing the flag that
1373 said "don't start kswapd" was a mistake. Let's hope we never revisit
1374 this mistake again - and certainly not this many times ;)
1375
1376 Acked-by: Mel Gorman <mgorman@suse.de>
1377 Acked-by: Johannes Weiner <hannes@cmpxchg.org>
1378 Cc: Rik van Riel <riel@redhat.com>
1379 Cc: Andrew Morton <akpm@linux-foundation.org>
1380 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1381commit 31f8d42d44b48ba72b586ca03e810cbbd21ea16b
1382Author: Linus Torvalds <torvalds@linux-foundation.org>
1383Date: Mon Dec 10 10:47:45 2012 -0800
1384
1385 Revert "mm: avoid waking kswapd for THP allocations when compaction is deferred or contended"
1386
1387 This reverts commit 782fd30406ecb9d9b082816abe0c6008fc72a7b0.
1388
1389 We are going to reinstate the __GFP_NO_KSWAPD flag that has been
1390 removed, the removal reverted, and then removed again. Making this
1391 commit a pointless fixup for a problem that was caused by the removal of
1392 __GFP_NO_KSWAPD flag.
1393
1394 The thing is, we really don't want to wake up kswapd for THP allocations
1395 (because they fail quite commonly under any kind of memory pressure,
1396 including when there is tons of memory free), and these patches were
1397 just trying to fix up the underlying bug: the original removal of
1398 __GFP_NO_KSWAPD in commit c654345924f7 ("mm: remove __GFP_NO_KSWAPD")
1399 was simply bogus.
1400
1401 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1402commit 37407ea7f93864c2cfc03edf8f37872ec539ea2b
1403Author: Linus Torvalds <torvalds@linux-foundation.org>
1404Date: Sun Sep 16 12:29:43 2012 -0700
1405
1406 Revert "sched: Improve scalability via 'CPU buddies', which withstand random perturbations"
1407
1408 This reverts commit 970e178985cadbca660feb02f4d2ee3a09f7fdda.
1409
1410 Nikolay Ulyanitsky reported thatthe 3.6-rc5 kernel has a 15-20%
1411 performance drop on PostgreSQL 9.2 on his machine (running "pgbench").
1412
1413 Borislav Petkov was able to reproduce this, and bisected it to this
1414 commit 970e178985ca ("sched: Improve scalability via 'CPU buddies' ...")
1415 apparently because the new single-idle-buddy model simply doesn't find
1416 idle CPU's to reschedule on aggressively enough.
1417
1418 Mike Galbraith suspects that it is likely due to the user-mode spinlocks
1419 in PostgreSQL not reacting well to preemption, but we don't really know
1420 the details - I'll just revert the commit for now.
1421
1422 There are hopefully other approaches to improve scheduler scalability
1423 without it causing these kinds of downsides.
1424
1425 Reported-by: Nikolay Ulyanitsky <lystor@gmail.com>
1426 Bisected-by: Borislav Petkov <bp@alien8.de>
1427 Acked-by: Mike Galbraith <efault@gmx.de>
1428 Cc: Andrew Morton <akpm@linux-foundation.org>
1429 Cc: Thomas Gleixner <tglx@linutronix.de>
1430 Cc: Ingo Molnar <mingo@kernel.org>
1431 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1432commit bc51b0c22cebf5c311a6f1895fcca9f78efd0478
1433Author: Linus Torvalds <torvalds@linux-foundation.org>
1434Date: Tue Jul 10 12:49:32 2012 -0700
1435
1436 Revert "of: match by compatible property first"
1437
1438 This reverts commit 107a84e61cdd3406c842a0e4be7efffd3a05dba6.
1439
1440 Meelis Roos reports a regression since 3.5-rc5 that stops Sun Fire V100
1441 and Sun Netra X1 sparc64 machines from booting, hanging after enabling
1442 serial console. He bisected it to commit 107a84e61cdd.
1443
1444 Rob Herring explains:
1445 "The problem is match combinations of compatible plus name and/or type
1446 fail to match correctly. I have a fix for this, but given how late it
1447 is for 3.5 I think it is best to revert this for now. There could be
1448 other cases that rely on the current although wrong behavior. I will
1449 post an updated version for 3.6."
1450
1451 Bisected-and-reported-by: Meelis Roos <mroos@linux.ee>
1452 Requested-by: Rob Herring <rob.herring@calxeda.com>
1453 Cc: Thierry Reding <thierry.reding@avionic-design.de>
1454 Cc: Grant Likely <grant.likely@secretlab.ca>
1455 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1456commit 8f53369b753f5f4c7684c2eb0b592152abb1dd00
1457Author: Linus Torvalds <torvalds@linux-foundation.org>
1458Date: Fri Jun 8 14:53:06 2012 -0700
1459
1460 Revert "drm/i915/crt: Do not rely upon the HPD presence pin"
1461
1462 This reverts commit 9e612a008fa7fe493a473454def56aa321479495.
1463
1464 It incorrectly finds VGA connectors where none are attached, apparently
1465 not noticing that nothing replied to the EDID queries, and happily using
1466 the default EDID modes that have nothing to do with actual hardware.
1467
1468 That in turn then causes X to fall down to the lowest common
1469 denominator, which is usually the default 1024x768 mode that is in the
1470 default EDID and pretty much anything supports).
1471
1472 I'd suggest that if not relying on the HDP pin, the code should at least
1473 check whether it gets valid EDID data back, rather than just assume
1474 there's something on the VGA connector.
1475
1476 Cc: Dave Airlie <airlied@linux.ie>
1477 Cc: Chris Wilson <chris@chris-wilson.co.uk>
1478 Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
1479 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1480commit 32ba9c3fcab960f0b0d332c86ebcd2c4870d9bb8
1481Author: Linus Torvalds <torvalds@linux-foundation.org>
1482Date: Fri Jun 8 10:34:03 2012 -0700
1483
1484 Revert "vfs: stop d_splice_alias creating directory aliases"
1485
1486 This reverts commit 7732a557b1342c6e6966efb5f07effcf99f56167 (and commit
1487 3f50fff4dace23d3cfeb195d5cd4ee813cee68b7, which was a follow-up
1488 cleanup).
1489
1490 We're chasing an elusive bug that Dave Jones can apparently reproduce
1491 using his system call fuzzer tool, and that looks like some kind of
1492 locking ordering problem on the directory i_mutex chain. Our i_mutex
1493 locking is rather complex, and depends on the topological ordering of
1494 the directories, which is why we have been very wary of splicing
1495 directory entries around.
1496
1497 Of course, we really don't want to ever see aliased unconnected
1498 directories anyway, so none of this should ever happen, but this revert
1499 aims to basically get us back to a known older state.
1500
1501 Bruce points to some of the previous discussion at
1502
1503 http://marc.info/?i=<20110310105821.GE22723@ZenIV.linux.org.uk>
1504
1505 and in particular a long post from Neil:
1506
1507 http://marc.info/?i=<20110311150749.2fa2be66@notabene.brown>
1508
1509 It should be noted that it's possible that Dave's problems come from
1510 other changes altohgether, including possibly just the fact that Dave
1511 constantly is teachning his fuzzer new tricks. So what appears to be a
1512 new bug could in fact be an old one that just gets newly triggered, but
1513 reverting these patches as "still under heavy discussion" is the right
1514 thing regardless.
1515
1516 Requested-by: Al Viro <viro@zeniv.linux.org.uk>
1517 Acked-by: J. Bruce Fields <bfields@fieldses.org>
1518 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1519commit 48d212a2eecaca2e1875925837ad27b2f43f48a3
1520Author: Linus Torvalds <torvalds@linux-foundation.org>
1521Date: Thu Jun 7 17:54:07 2012 -0700
1522
1523 Revert "mm: correctly synchronize rss-counters at exit/exec"
1524
1525 This reverts commit 40af1bbdca47e5c8a2044039bb78ca8fd8b20f94.
1526
1527 It's horribly and utterly broken for at least the following reasons:
1528
1529 - calling sync_mm_rss() from mmput() is fundamentally wrong, because
1530 there's absolutely no reason to believe that the task that does the
1531 mmput() always does it on its own VM. Example: fork, ptrace, /proc -
1532 you name it.
1533
1534 - calling it *after* having done mmdrop() on it is doubly insane, since
1535 the mm struct may well be gone now.
1536
1537 - testing mm against NULL before you call it is insane too, since a
1538 NULL mm there would have caused oopses long before.
1539
1540 .. and those are just the three bugs I found before I decided to give up
1541 looking for me and revert it asap. I should have caught it before I
1542 even took it, but I trusted Andrew too much.
1543
1544 Cc: Konstantin Khlebnikov <khlebnikov@openvz.org>
1545 Cc: Markus Trippelsdorf <markus@trippelsdorf.de>
1546 Cc: Hugh Dickins <hughd@google.com>
1547 Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
1548 Cc: Oleg Nesterov <oleg@redhat.com>
1549 Cc: Andrew Morton <akpm@linux-foundation.org>
1550 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1551commit 68e3e92620c323703bc7db75c2ba15239ee85c39
1552Author: Linus Torvalds <torvalds@linux-foundation.org>
1553Date: Sun Jun 3 20:05:57 2012 -0700
1554
1555 Revert "mm: compaction: handle incorrect MIGRATE_UNMOVABLE type pageblocks"
1556
1557 This reverts commit 5ceb9ce6fe9462a298bb2cd5c9f1ca6cb80a0199.
1558
1559 That commit seems to be the cause of the mm compation list corruption
1560 issues that Dave Jones reported. The locking (or rather, absense
1561 there-of) is dubious, as is the use of the 'page' variable once it has
1562 been found to be outside the pageblock range.
1563
1564 So revert it for now, we can re-visit this for 3.6. If we even need to:
1565 as Minchan Kim says, "The patch wasn't a bug fix and even test workload
1566 was very theoretical".
1567
1568 Reported-and-tested-by: Dave Jones <davej@redhat.com>
1569 Acked-by: Hugh Dickins <hughd@google.com>
1570 Acked-by: KOSAKI Motohiro <kosaki.motohiro@gmail.com>
1571 Acked-by: Minchan Kim <minchan@kernel.org>
1572 Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
1573 Cc: Kyungmin Park <kyungmin.park@samsung.com>
1574 Cc: Andrew Morton <akpm@linux-foundation.org>
1575 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1576commit 2e321806b681b1920b6dfa7d81bbe3d312fe1d19
1577Author: Linus Torvalds <torvalds@linux-foundation.org>
1578Date: Mon May 21 18:48:10 2012 -0700
1579
1580 Revert "vfs: remove unnecessary d_unhashed() check from __d_lookup_rcu"
1581
1582 This reverts commit 8c01a529b861ba97c7d78368e6a5d4d42e946f75.
1583
1584 It turns out the d_unhashed() check isn't unnecessary after all: while
1585 it's true that unhashing will increment the sequence numbers, that does
1586 not necessarily invalidate the RCU lookup, because it might have seen
1587 the dentry pointer (before it got unhashed), but by the time it loaded
1588 the sequence number, it could have seen the *new* sequence number (after
1589 it got unhashed).
1590
1591 End result: we might look up an unhashed dentry that is about to be
1592 freed, with the sequence number never indicating anything bad about it.
1593 So checking that the dentry is still hashed (*after* reading the sequence
1594 number) is indeed the proper fix, and was never unnecessary.
1595
1596 Reported-by: Dave Jones <davej@redhat.com>
1597 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1598commit fcbf94b9dedd2ce08e798a99aafc94fec8668161
1599Author: Linus Torvalds <torvalds@linux-foundation.org>
1600Date: Sat Apr 28 08:29:56 2012 -0700
1601
1602 Revert "autofs: work around unhappy compat problem on x86-64"
1603
1604 This reverts commit a32744d4abae24572eff7269bc17895c41bd0085.
1605
1606 While that commit was technically the right thing to do, and made the
1607 x86-64 compat mode work identically to native 32-bit mode (and thus
1608 fixing the problem with a 32-bit systemd install on a 64-bit kernel), it
1609 turns out that the automount binaries had workarounds for this compat
1610 problem.
1611
1612 Now, the workarounds are disgusting: doing an "uname()" to find out the
1613 architecture of the kernel, and then comparing it for the 64-bit cases
1614 and fixing up the size of the read() in automount for those. And they
1615 were confused: it's not actually a generic 64-bit issue at all, it's
1616 very much tied to just x86-64, which has different alignment for an
1617 'u64' in 64-bit mode than in 32-bit mode.
1618
1619 But the end result is that fixing the compat layer actually breaks the
1620 case of a 32-bit automount on a x86-64 kernel.
1621
1622 There are various approaches to fix this (including just doing a
1623 "strcmp()" on current->comm and comparing it to "automount"), but I
1624 think that I will do the one that teaches pipes about a special "packet
1625 mode", which will allow user space to not have to care too deeply about
1626 the padding at the end of the autofs packet.
1627
1628 That change will make the compat workaround unnecessary, so let's revert
1629 it first, and get automount working again in compat mode. The
1630 packetized pipes will then fix autofs for systemd.
1631
1632 Reported-and-requested-by: Michael Tokarev <mjt@tls.msk.ru>
1633 Cc: Ian Kent <raven@themaw.net>
1634 Cc: stable@kernel.org # for 3.3
1635 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1636commit 19244ad06b70ed84931df868583547ce1cd3a186
1637Author: Linus Torvalds <torvalds@linux-foundation.org>
1638Date: Fri Apr 20 11:19:35 2012 -0700
1639
1640 Revert "ACPI: ignore FADT reset-reg-sup flag"
1641
1642 This reverts commit cf450136bfde77c7f95065c91bffded4aa7fa731.
1643
1644 It breaks reboot on at least one Thinkpad T43, as reported by Jörg Otte:
1645 "On reboot it shuts down as normal.
1646 The last lines displayed are:
1647
1648 >Unmounting temporary filesystems.. [OK]
1649 >Deactivating swap... [OK]
1650 >Unmounting local filesystems... [OK]
1651 >Will now restart
1652 > Restarting system
1653
1654 Then I hear it accessing the cd-drive, but then it's being stuck."
1655
1656 Jörg bisected the regression to this commit.
1657
1658 That commit fixes another machine (see
1659
1660 https://bugzilla.kernel.org/show_bug.cgi?id=11533
1661
1662 for details) that has a BIOS bug and doesn't support ACPI reset.
1663 However, at least one of those other reporters no longer even has the
1664 machine in question, and had a different workaround to begin with.
1665 Besides, it clearly was a buggy BIOS. Let's not break the correct case
1666 to fix that case.
1667
1668 Reported-and-bisected-by: Jörg Otte <jrg.otte@googlemail.com>
1669 Cc: linux-acpi@vger.kernel.org
1670 Cc: Len Brown <lenb@kernel.org>
1671 Cc: Peter Anvin <hpa@zytor.com>
1672 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1673commit f3718a818f7fb3636130d5f34bde8df34f45c5e5
1674Author: Linus Torvalds <torvalds@linux-foundation.org>
1675Date: Wed Apr 4 08:16:25 2012 -0700
1676
1677 Revert "nouveau/bios: Fix tracking of BIOS image data"
1678
1679 This reverts commit d06221c0617ab6d0bc41c4980cefdd9c8cc9a1c1.
1680
1681 It turns out to trigger the "BUG_ON(!PageCompound(page))" in kfree(),
1682 apparently because the code ends up trying to free somethng that was
1683 never kmalloced in the first place.
1684
1685 BenH points out that the patch was untested and wasn't meant to go into
1686 the upstream kernel that quickly in the first place.
1687
1688 Backtrace:
1689 bios_shadow
1690 bios_shadow_prom
1691 nv_mask
1692 init_io
1693 bios_shadow
1694 nouveau_bios_init
1695 NVReadVgaCrtc
1696 NVSetOwner
1697 nouveau_card_init
1698 nouveau_load
1699
1700 Reported-by: Meelis Roos <mroos@linux.ee>
1701 Requested-by: Dave Airlie <airlied@gmail.com>
1702 Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
1703 Cc: Ben Skeggs <bskeggs@redhat.com>
1704 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1705commit 6268b325c3066234e7bddb99d2b98bcedb0c0033
1706Author: Linus Torvalds <torvalds@linux-foundation.org>
1707Date: Thu Mar 29 17:00:56 2012 -0700
1708
1709 Revert "ext4: don't release page refs in ext4_end_bio()"
1710
1711 This reverts commit b43d17f319f2c502b17139d1cf70731b2b62c644.
1712
1713 Dave Jones reports that it causes lockups on his laptop, and his debug
1714 output showed a lot of processes hung waiting for page_writeback (or
1715 more commonly - processes hung waiting for a lock that was held during
1716 that writeback wait).
1717
1718 The page_writeback hint made Ted suggest that Dave look at this commit,
1719 and Dave verified that reverting it makes his problems go away.
1720
1721 Ted says:
1722 "That commit fixes a race which is seen when you write into fallocated
1723 (and hence uninitialized) disk blocks under *very* heavy memory
1724 pressure. Furthermore, although theoretically it could trigger under
1725 normal direct I/O writes, it only seems to trigger if you are issuing
1726 a huge number of AIO writes, such that a just-written page can get
1727 evicted from memory, and then read back into memory, before the
1728 workqueue has a chance to update the extent tree.
1729
1730 This race has been around for a little over a year, and no one noticed
1731 until two months ago; it only happens under fairly exotic conditions,
1732 and in fact even after trying very hard to create a simple repro under
1733 lab conditions, we could only reproduce the problem and confirm the
1734 fix on production servers running MySQL on very fast PCIe-attached
1735 flash devices.
1736
1737 Given that Dave was able to hit this problem pretty quickly, if we
1738 confirm that this commit is at fault, the only reasonable thing to do
1739 is to revert it IMO."
1740
1741 Reported-and-tested-by: Dave Jones <davej@redhat.com>
1742 Acked-by: Theodore Ts'o <tytso@mit.edu>
1743 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1744commit 4293f20c19f44ca66e5ac836b411d25e14b9f185
1745Author: Linus Torvalds <torvalds@linux-foundation.org>
1746Date: Wed Mar 7 08:21:19 2012 -0800
1747
1748 Revert "CPU hotplug, cpusets, suspend: Don't touch cpusets during suspend/resume"
1749
1750 This reverts commit 8f2f748b0656257153bcf0941df8d6060acc5ca6.
1751
1752 It causes some odd regression that we have not figured out, and it's too
1753 late in the -rc series to try to figure it out now.
1754
1755 As reported by Konstantin Khlebnikov, it causes consistent hangs on his
1756 laptop (Thinkpad x220: 2x cores + HT). They can be avoided by adding
1757 calls to "rebuild_sched_domains();" in cpuset_cpu_[in]active() for the
1758 CPU_{ONLINE/DOWN_FAILED/DOWN_PREPARE}_FROZEN cases, but it's not at all
1759 clear why, and it makes no sense.
1760
1761 Konstantin's config doesn't even have CONFIG_CPUSETS enabled, just to
1762 make things even more interesting. So it's not the cpusets, it's just
1763 the scheduling domains.
1764
1765 So until this is understood, revert.
1766
1767 Bisected-reported-and-tested-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
1768 Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
1769 Acked-by: Ingo Molnar <mingo@elte.hu>
1770 Acked-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
1771 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1772commit 951880e634a79884236a575b896abf55c39ae0bf
1773Author: Linus Torvalds <torvalds@linux-foundation.org>
1774Date: Tue Jan 17 10:19:41 2012 -0800
1775
1776 Revert "capabitlies: ns_capable can use the cap helpers rather than lsm call"
1777
1778 This reverts commit d2a7009f0bb03fa22ad08dd25472efa0568126b9.
1779
1780 J. R. Okajima explains:
1781
1782 "After this commit, I am afraid access(2) on NFS may not work
1783 correctly. The scenario based upon my guess.
1784 - access(2) overrides the credentials.
1785 - calls inode_permission() -- ... -- generic_permission() --
1786 ns_capable().
1787 - while the old ns_capable() calls security_capable(current_cred()),
1788 the new ns_capable() calls has_ns_capability(current) --
1789 security_capable(__task_cred(t)).
1790
1791 current_cred() returns current->cred which is effective (overridden)
1792 credentials, but __task_cred(current) returns current->real_cred (the
1793 NFSD's credential). And the overridden credentials by access(2) lost."
1794
1795 Requested-by: J. R. Okajima <hooanon05@yahoo.co.jp>
1796 Acked-by: Eric Paris <eparis@redhat.com>
1797 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1798commit f423fc627b05f47bc9305f9661630fce30f208f9
1799Author: Linus Torvalds <torvalds@linux-foundation.org>
1800Date: Wed Jan 4 07:57:22 2012 -0800
1801
1802 Revert "rtc: Expire alarms after the time is set."
1803
1804 This reverts commit 93b2ec0128c431148b216b8f7337c1a52131ef03.
1805
1806 The call to "schedule_work()" in rtc_initialize_alarm() happens too
1807 early, and can cause oopses at bootup
1808
1809 Neil Brown explains why we do it:
1810
1811 "If you set an alarm in the future, then shutdown and boot again after
1812 that time, then you will end up with a timer_queue node which is in
1813 the past.
1814
1815 When this happens the queue gets stuck. That entry-in-the-past won't
1816 get removed until and interrupt happens and an interrupt won't happen
1817 because the RTC only triggers an interrupt when the alarm is "now".
1818
1819 So you'll find that e.g. "hwclock" will always tell you that
1820 'select' timed out.
1821
1822 So we force the interrupt work to happen at the start just in case."
1823
1824 and has a patch that convert it to do things in-process rather than with
1825 the worker thread, but right now it's too late to play around with this,
1826 so we just revert the patch that caused problems for now.
1827
1828 Reported-by: Sander Eikelenboom <linux@eikelenboom.it>
1829 Requested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
1830 Requested-by: John Stultz <john.stultz@linaro.org>
1831 Cc: Neil Brown <neilb@suse.de>
1832 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1833commit 157e8bf8b4823bfcdefa6c1548002374b61f61df
1834Author: Linus Torvalds <torvalds@linux-foundation.org>
1835Date: Tue Jan 3 17:32:13 2012 -0800
1836
1837 Revert "rtc: Disable the alarm in the hardware"
1838
1839 This reverts commit c0afabd3d553c521e003779c127143ffde55a16f.
1840
1841 It causes failures on Toshiba laptops - instead of disabling the alarm,
1842 it actually seems to enable it on the affected laptops, resulting in
1843 (for example) the laptop powering on automatically five minutes after
1844 shutdown.
1845
1846 There's a patch for it that appears to work for at least some people,
1847 but it's too late to play around with this, so revert for now and try
1848 again in the next merge window.
1849
1850 See for example
1851
1852 http://bugs.debian.org/652869
1853
1854 Reported-and-bisected-by: Andreas Friedrich <afrie@gmx.net> (Toshiba Tecra)
1855 Reported-by: Antonio-M. Corbi Bellot <antonio.corbi@ua.es> (Toshiba Portege R500)
1856 Reported-by: Marco Santos <marco.santos@waynext.com> (Toshiba Portege Z830)
1857 Reported-by: Christophe Vu-Brugier <cvubrugier@yahoo.fr> (Toshiba Portege R830)
1858 Cc: Jonathan Nieder <jrnieder@gmail.com>
1859 Requested-by: John Stultz <john.stultz@linaro.org>
1860 Cc: stable@kernel.org # for the versions that applied this
1861 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1862commit 3b87487ac5008072f138953b07505a7e3493327f
1863Author: Linus Torvalds <torvalds@linux-foundation.org>
1864Date: Fri Dec 30 13:24:40 2011 -0800
1865
1866 Revert "clockevents: Set noop handler in clockevents_exchange_device()"
1867
1868 This reverts commit de28f25e8244c7353abed8de0c7792f5f883588c.
1869
1870 It results in resume problems for various people. See for example
1871
1872 http://thread.gmane.org/gmane.linux.kernel/1233033
1873 http://thread.gmane.org/gmane.linux.kernel/1233389
1874 http://thread.gmane.org/gmane.linux.kernel/1233159
1875 http://thread.gmane.org/gmane.linux.kernel/1227868/focus=1230877
1876
1877 and the fedora and ubuntu bug reports
1878
1879 https://bugzilla.redhat.com/show_bug.cgi?id=767248
1880 https://bugs.launchpad.net/ubuntu/+source/linux/+bug/904569
1881
1882 which got bisected down to the stable version of this commit.
1883
1884 Reported-by: Jonathan Nieder <jrnieder@gmail.com>
1885 Reported-by: Phil Miller <mille121@illinois.edu>
1886 Reported-by: Philip Langdale <philipl@overt.org>
1887 Reported-by: Tim Gardner <tim.gardner@canonical.com>
1888 Cc: Thomas Gleixner <tglx@linutronix.de>
1889 Cc: Greg KH <gregkh@suse.de>
1890 Cc: stable@kernel.org # for stable kernels that applied the original
1891 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1892commit ed4a51842a9d9e618d4f4c31349b15b974dba5df
1893Author: Linus Torvalds <torvalds@linux-foundation.org>
1894Date: Fri Dec 16 12:58:39 2011 -0800
1895
1896 Revert "drm/i915: fix infinite recursion on unbind due to ilk vt-d w/a"
1897
1898 This reverts commit eb1711bb94991e93669c5a1b5f84f11be2d51ea1.
1899
1900 It blows up the i915 seqno tracking, resulting in the
1901
1902 BUG_ON(seqno == 0);
1903
1904 in i915_wait_request() triggering, which will cause lock-ups.
1905
1906 See for example
1907 https://bugs.launchpad.net/ubuntu/+source/linux/+bug/903010
1908 https://lkml.org/lkml/2011/12/14/395
1909
1910 Reported-requested-and-tested-by: Dirk Hohndel <dirk@hohndel.org>
1911 Reported-by: Richard Eames <Richard.Eames@flinders.edu.au>
1912 Reported-by: Rocko Requin <rockorequin@hotmail.com>
1913 Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
1914 Cc: Dave Airlie <airlied@redhat.com>
1915 Cc: Chris Wilson <chris@chris-wilson.co.uk>
1916 Cc: Keith Packard <keithp@keithp.com>
1917 Cc: Eric Anholt <eric@anholt.net>
1918 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1919commit b4bbb02934e4511d9083f15c23e90703482e84ad
1920Author: Linus Torvalds <torvalds@linux-foundation.org>
1921Date: Tue Nov 22 15:09:20 2011 -0800
1922
1923 Revert "of/irq: of_irq_find_parent: check for parent equal to child"
1924
1925 This reverts commit dc9372808412edbc653a675a526c2ee6c0c14a91.
1926
1927 As requested by Ben Herrenschmidt:
1928 "This breaks some powerpc platforms at least. The practice of having
1929 a node provide an explicit "interrupt-parent" property pointing to
1930 itself is an old trick that we've used in the past to allow a
1931 device-node to have interrupts routed to different controllers.
1932
1933 In that case, the node also contains an interrupt-map, so the node is
1934 its own parent, the interrupt resolution hits the map, which then can
1935 route each individual interrupt to a different parent."
1936
1937 Grant says:
1938 "Ah, nuts, yes that is broken then. Yes, please revert the commit and
1939 Rob & I will come up with a better solution.
1940
1941 Rob, I think it can be done by explicitly checking for np ==
1942 desc->interrupt_parent in of_irq_init() instead of relying on
1943 of_irq_find_parent() returning NULL."
1944
1945 Requested-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
1946 Acked-by: Grant Likely <grant.likely@secretlab.ca>
1947 Cc: Rob Herring <rob.herring@calxeda.com>
1948 Cc: devicetree-discuss@lists.ozlabs.org
1949 Cc: linuxppc-dev <linuxppc-dev@lists.ozlabs.org>
1950 Cc: Tanmay Inamdar <tinamdar@apm.com>
1951 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1952commit 5e442a493fc59fa536c76db1fff5b49ca36a88c5
1953Author: Linus Torvalds <torvalds@linux-foundation.org>
1954Date: Wed Nov 9 18:16:00 2011 -0500
1955
1956 Revert "proc: fix races against execve() of /proc/PID/fd**"
1957
1958 This reverts commit aa6afca5bcaba8101f3ea09d5c3e4100b2b9f0e5.
1959
1960 It escalates of some of the google-chrome SELinux problems with ptrace
1961 ("Check failed: pid_ > 0. Did not find zygote process"), and Andrew
1962 says that it is also causing mystery lockdep reports.
1963
1964 Reported-by: Alex VillacÃs Lasso <a_villacis@palosanto.com>
1965 Requested-by: James Morris <jmorris@namei.org>
1966 Requested-by: Andrew Morton <akpm@linux-foundation.org>
1967 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1968commit 31555213f03bca37d2c02e10946296052f4ecfcd
1969Author: Linus Torvalds <torvalds@linux-foundation.org>
1970Date: Sun Nov 6 22:22:16 2011 -0800
1971
1972 Revert "hvc_console: display printk messages on console."
1973
1974 This reverts commit 361162459f62dc0826b82c9690a741a940f457f0.
1975
1976 It causes an infinite loop when booting Linux under Xen, as so:
1977
1978 [ 2.382984] console [hvc0] enabled
1979 [ 2.382984] console [hvc0] enabled
1980 [ 2.382984] console [hvc0] enabled
1981 ...
1982
1983 as reported by Konrad Rzeszutek Wilk. And Rusty reports the same for
1984 lguest. He goes on to say:
1985
1986 "This is not a concurrency problem: the issue seems to be that
1987 calling register_console() twice on the same struct console is a bad
1988 idea."
1989
1990 and Greg says he'll fix it up properly at some point later. Revert for now.
1991
1992 Reported-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
1993 Reported-by: Rusty Russell <rusty@ozlabs.org>
1994 Requested-by: Stephen Rothwell <sfr@canb.auug.org.au>
1995 Cc: Miche Baker-Harvey <miche@google.com>
1996 Cc: Greg Kroah-Hartman <gregkh@suse.de>
1997 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1998commit 4536e4d1d21c8172402a2217b0fa1880665ace36
1999Author: Linus Torvalds <torvalds@linux-foundation.org>
2000Date: Thu Nov 3 07:44:04 2011 -0700
2001
2002 Revert "perf: Add PM notifiers to fix CPU hotplug races"
2003
2004 This reverts commit 144060fee07e9c22e179d00819c83c86fbcbf82c.
2005
2006 It causes a resume regression for Andi on his Acer Aspire 1830T post
2007 3.1. The screen just stays black after wakeup.
2008
2009 Also, it really looks like the wrong way to suspend and resume perf
2010 events: I think they should be done as part of the CPU suspend and
2011 resume, rather than as a notifier that does smp_call_function().
2012
2013 Reported-by: Andi Kleen <andi@firstfloor.org>
2014 Acked-by: Ingo Molnar <mingo@elte.hu>
2015 Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
2016 Cc: Rafael J. Wysocki <rjw@sisk.pl>
2017 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018commit 69dd3d8e29e294caaf63eb5e8a72d250279f9e5f
2019Author: Linus Torvalds <torvalds@linux-foundation.org>
2020Date: Tue Aug 23 10:36:51 2011 -0700
2021
2022 Revert "irq: Always set IRQF_ONESHOT if no primary handler is specified"
2023
2024 This reverts commit f3637a5f2e2eb391ff5757bc83fb5de8f9726464.
2025
2026 It turns out that this breaks several drivers, one example being OMAP
2027 boards which use the on-board OMAP UARTs and the omap-serial driver that
2028 will not boot to userspace after the commit.
2029
2030 Paul Walmsley reports that enabling CONFIG_DEBUG_SHIRQ reveals 'IRQ
2031 handler type mismatch' errors:
2032
2033 IRQ handler type mismatch for IRQ 74
2034 current handler: serial idle
2035 ...
2036
2037 and the reason is that setting IRQF_ONESHOT will now result in those
2038 interrupt handlers having different IRQF flags, and thus being
2039 unsharable. So the commit log in the reverted commit:
2040
2041 "Since it is required for those users and
2042 there is no difference for others it makes sense to add this flag
2043 unconditionally."
2044
2045 is simply not true: there may not be any difference from a "actions at
2046 irq time", but there is a *big* difference wrt this flag testing irq
2047 management (see __setup_irq() in kernel/irq/manage.c).
2048
2049 One solution may be to stop verifying IRQF_ONESHOT in __setup_irq(), but
2050 right now the safe course of action is to revert the change. Let's
2051 revisit this in a later merge window.
2052
2053 Reported-by: Paul Walmsley <paul@pwsan.com>
2054 Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
2055 Requested-by: Alan Cox <alan@lxorguk.ukuu.org.uk>
2056 Acked-by: Thomas Gleixner <tglx@linutronix.de>
2057 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2058commit a9f729f0e28bb4e4ab0d9e9e3c1675fe4b910f47
2059Author: Linus Torvalds <torvalds@linux-foundation.org>
2060Date: Thu Aug 11 08:58:41 2011 -0700
2061
2062 Revert "EDAC: Correct Kconfig dependencies"
2063
2064 This reverts commit af9d220bac41dc3201893e1601cc7c44f7da4498.
2065
2066 It turns out that one was meant to be applied on top of the edac.git
2067 tree in -next that has more i7core_edac changes, but that wasn't clear
2068 in the original email.
2069
2070 Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
2071 Acked-by: Borislav Petkov <borislav.petkov@amd.com>
2072 Cc: Randy Dunlap <rdunlap@xenotime.net>
2073 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2074commit 13fca640bb8ab611a50e0ba120b186faa2994d6c
2075Author: Linus Torvalds <torvalds@linux-foundation.org>
2076Date: Wed Jun 15 21:53:52 2011 -0700
2077
2078 Revert "fs/exec.c: use BUILD_BUG_ON for VM_STACK_FLAGS & VM_STACK_INCOMPLETE_SETUP"
2079
2080 This reverts commit 7f81c8890c15a10f5220bebae3b6dfae4961962a.
2081
2082 It turns out that it's not actually a build-time check on x86-64 UML,
2083 which does some seriously crazy stuff with VM_STACK_FLAGS.
2084
2085 The VM_STACK_FLAGS define depends on the arch-supplied
2086 VM_STACK_DEFAULT_FLAGS value, and on x86-64 UML we have
2087
2088 arch/um/sys-x86_64/shared/sysdep/vm-flags.h:
2089
2090 #define VM_STACK_DEFAULT_FLAGS \
2091 (test_thread_flag(TIF_IA32) ? vm_stack_flags32 : vm_stack_flags)
2092
2093 #define VM_STACK_DEFAULT_FLAGS vm_stack_flags
2094
2095 (yes, seriously: two different #define's for that thing, with the first
2096 one being inside an "#ifdef TIF_IA32")
2097
2098 It's possible that it is UML that should just be fixed in this area, but
2099 for now let's just undo the (very small) optimization.
2100
2101 Reported-by: Randy Dunlap <randy.dunlap@oracle.com>
2102 Acked-by: Andrew Morton <akpm@linux-foundation.org>
2103 Cc: Michal Hocko <mhocko@suse.cz>
2104 Cc: Richard Weinberger <richard@nod.at>
2105 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2106commit bb3d6bf1919a20c56b3257b4ec09e67a9226cfb2
2107Author: Linus Torvalds <torvalds@linux-foundation.org>
2108Date: Sat Jun 4 07:00:50 2011 +0900
2109
2110 Revert "ASoC: Update cx20442 for TTY API change"
2111
2112 This reverts commit ed0bd2333cffc3d856db9beb829543c1dfc00982.
2113
2114 Since we reverted the TTY API change, we should revert the ASoC update
2115 to it too.
2116
2117 Cc: Mark Brown <broonie@opensource.wolfsonmicro.com>
2118 Cc: Liam Girdwood <lrg@ti.com>
2119 Cc: Greg Kroah-Hartman <gregkh@suse.de>
2120 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2121commit 55db4c64eddf37e31279ec15fe90314713bc9cfa
2122Author: Linus Torvalds <torvalds@linux-foundation.org>
2123Date: Sat Jun 4 06:33:24 2011 +0900
2124
2125 Revert "tty: make receive_buf() return the amout of bytes received"
2126
2127 This reverts commit b1c43f82c5aa265442f82dba31ce985ebb7aa71c.
2128
2129 It was broken in so many ways, and results in random odd pty issues.
2130
2131 It re-introduced the buggy schedule_work() in flush_to_ldisc() that can
2132 cause endless work-loops (see commit a5660b41af6a: "tty: fix endless
2133 work loop when the buffer fills up").
2134
2135 It also used an "unsigned int" return value fo the ->receive_buf()
2136 function, but then made multiple functions return a negative error code,
2137 and didn't actually check for the error in the caller.
2138
2139 And it didn't actually work at all. BenH bisected down odd tty behavior
2140 to it:
2141 "It looks like the patch is causing some major malfunctions of the X
2142 server for me, possibly related to PTYs. For example, cat'ing a
2143 large file in a gnome terminal hangs the kernel for -minutes- in a
2144 loop of what looks like flush_to_ldisc/workqueue code, (some ftrace
2145 data in the quoted bits further down).
2146
2147 ...
2148
2149 Some more data: It -looks- like what happens is that the
2150 flush_to_ldisc work queue entry constantly re-queues itself (because
2151 the PTY is full ?) and the workqueue thread will basically loop
2152 forver calling it without ever scheduling, thus starving the consumer
2153 process that could have emptied the PTY."
2154
2155 which is pretty much exactly the problem we fixed in a5660b41af6a.
2156
2157 Milton Miller pointed out the 'unsigned int' issue.
2158
2159 Reported-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2160 Reported-by: Milton Miller <miltonm@bga.com>
2161 Cc: Stefan Bigler <stefan.bigler@keymile.com>
2162 Cc: Toby Gray <toby.gray@realvnc.com>
2163 Cc: Felipe Balbi <balbi@ti.com>
2164 Cc: Greg Kroah-Hartman <gregkh@suse.de>
2165 Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
2166 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2167commit 1fa7b6a29c61358cc2ca6f64cef4aa0e1a7ca74c
2168Author: Linus Torvalds <torvalds@linux-foundation.org>
2169Date: Thu Jun 2 06:11:24 2011 +0900
2170
2171 Revert "mm: fail GFP_DMA allocations when ZONE_DMA is not configured"
2172
2173 This reverts commit a197b59ae6e8bee56fcef37ea2482dc08414e2ac.
2174
2175 As rmk says:
2176 "Commit a197b59ae6e8 (mm: fail GFP_DMA allocations when ZONE_DMA is not
2177 configured) is causing regressions on ARM with various drivers which
2178 use GFP_DMA.
2179
2180 The behaviour up until now has been to silently ignore that flag when
2181 CONFIG_ZONE_DMA is not enabled, and to allocate from the normal zone.
2182 However, as a result of the above commit, such allocations now fail
2183 which causes drivers to fail. These are regressions compared to the
2184 previous kernel version."
2185
2186 so just revert it.
2187
2188 Requested-by: Russell King <linux@arm.linux.org.uk>
2189 Acked-by: Andrew Morton <akpm@linux-foundation.org>
2190 Cc: David Rientjes <rientjes@google.com>
2191 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2192commit 982b2035d9d7033f63db187bac55e9d8998b0266
2193Author: Linus Torvalds <torvalds@linux-foundation.org>
2194Date: Thu May 12 12:19:43 2011 -0700
2195
2196 Revert "drm/i915: Only enable the plane after setting the fb base (pre-ILK)"
2197
2198 This reverts commit 49183b2818de6899383bb82bc032f9344d6791ff.
2199
2200 Quoth Franz Melchior:
2201
2202 "This patch introduces a bug on my infamous "Acer Travelmate
2203 5735Z-452G32Mnss": when KMS takes over, the frame buffer contents get
2204 completely garbled up on screen, with colored stripes and unreadable
2205 text (photo on request). Only when X11 is started, the screen gets
2206 restored again. Closing and re-opening the lid partly cures the
2207 mess, too: it makes the font readable, though horizontally stretched."
2208
2209 Acked-by: Keith Packard <keithp@keithp.com>
2210 Cc: Chris Wilson <chris@chris-wilson.co.uk>
2211 Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
2212 Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
2213 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2214commit ce8453776d68982cfe93bcb28191af8ccad01f45
2215Author: Linus Torvalds <torvalds@linux-foundation.org>
2216Date: Wed May 11 18:58:16 2011 -0700
2217
2218 Revert "Bluetooth: fix shutdown on SCO sockets"
2219
2220 This reverts commit f21ca5fff6e548833fa5ee8867239a8378623150.
2221
2222 Quoth Gustavo F. Padovan:
2223 "Commit f21ca5fff6e548833fa5ee8867239a8378623150 can cause a NULL
2224 dereference if we call shutdown in a bluetooth SCO socket and doesn't
2225 wait the shutdown completion to call close(). Please revert it. I
2226 may have a fix for it soon, but we don't have time anymore, so revert
2227 is the way to go. ;)"
2228
2229 Requested-by: Gustavo F. Padovan <padovan@profusion.mobi>
2230 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2231commit 71460af58f8565110160283849db4d6bf7e1efa1
2232Author: Linus Torvalds <torvalds@linux-foundation.org>
2233Date: Tue Apr 19 10:54:44 2011 -0700
2234
2235 Revert "[media] V4L: videobuf, don't use dma addr as physical"
2236
2237 This reverts commit 35d9f510b67b10338161aba6229d4f55b4000f5b.
2238
2239 Quoth Jiri Slaby:
2240 "It fixes mmap when IOMMU is used on x86 only, but breaks architectures
2241 like ARM or PPC where virt_to_phys(dma_alloc_coherent) doesn't work.
2242 We need there dma_mmap_coherent or similar (the trickery what
2243 snd_pcm_default_mmap does but in some saner way). But this cannot be
2244 done at this phase."
2245
2246 Requested-by: Jiri Slaby <jslaby@suse.cz>
2247 Cc: Russell King - ARM Linux <linux@arm.linux.org.uk>
2248 Cc: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
2249 Acked-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2250 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2251commit be85bccaa5aa5a11dcaf85f9e945ffefd253f631
2252Author: Linus Torvalds <torvalds@linux-foundation.org>
2253Date: Tue Apr 12 13:35:56 2011 -0700
2254
2255 Revert "vfs: Export file system uuid via /proc/<pid>/mountinfo"
2256
2257 This reverts commit 93f1c20bc8cdb757be50566eff88d65c3b26881f.
2258
2259 It turns out that libmount misparses it because it adds a '-' character
2260 in the uuid string, which libmount then incorrectly confuses with the
2261 separator string (" - ") at the end of all the optional arguments.
2262
2263 Upstream libmount (in the util-linux tree) has been fixed, but until
2264 that fix actually percolates up to users, we'd better not expose this
2265 change in the kernel.
2266
2267 Let's revisit this later (possibly by exposing the UUID without any '-'
2268 characters in it, avoiding the user-space bug).
2269
2270 Reported-by: Dave Jones <davej@redhat.com>
2271 Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
2272 Cc: Al Viro <viro@zeniv.linux.org.uk>
2273 Cc: Karel Zak <kzak@redhat.com>
2274 Cc: Ram Pai <linuxram@us.ibm.com>
2275 Cc: Miklos Szeredi <mszeredi@suse.cz>
2276 Cc: Eric Sandeen <sandeen@redhat.com>
2277 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2278commit 52d3c03675fdbe1965b9b1909072b40ad2f80063
2279Author: Linus Torvalds <torvalds@linux-foundation.org>
2280Date: Mon Mar 14 15:17:07 2011 -0700
2281
2282 Revert "oom: oom_kill_process: fix the child_points logic"
2283
2284 This reverts the parent commit. I hate doing that, but it's generating
2285 some discussion ("half of it is right"), and since I am planning on
2286 doing the 2.6.38 release later today we can punt it to stable if
2287 required. Let's not rock the boat right now.
2288
2289 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2290commit 8d1dc20e8d689c7e6a0a4d2c94e36a99d5793ecb
2291Author: Linus Torvalds <torvalds@linux-foundation.org>
2292Date: Tue Mar 1 13:23:27 2011 -0800
2293
2294 Revert "TPM: Long default timeout fix"
2295
2296 This reverts commit c4ff4b829ef9e6353c0b133b7adb564a68054979.
2297
2298 Ted Ts'o reports:
2299
2300 "TPM is working for me so I can log into employer's network in 2.6.37.
2301 It broke when I tried 2.6.38-rc6, with the following relevant lines
2302 from my dmesg:
2303
2304 [ 11.081627] tpm_tis 00:0b: 1.2 TPM (device-id 0x0, rev-id 78)
2305 [ 25.734114] tpm_tis 00:0b: Operation Timed out
2306 [ 78.040949] tpm_tis 00:0b: Operation Timed out
2307
2308 This caused me to get suspicious, especially since the _other_ TPM
2309 commit in 2.6.38 had already been reverted, so I tried reverting
2310 commit c4ff4b829e: "TPM: Long default timeout fix". With this commit
2311 reverted, my TPM on my Lenovo T410 is once again working."
2312
2313 Requested-and-tested-by: Theodore Ts'o <tytso@mit.edu>
2314 Acked-by: Rajiv Andrade <srajiv@linux.vnet.ibm.com>
2315 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2316commit 78794b2cdeac37ac1fd950fc9c4454b56d88ac03
2317Author: Linus Torvalds <torvalds@linux-foundation.org>
2318Date: Wed Feb 23 19:42:03 2011 -0800
2319
2320 Revert "Bluetooth: Enable USB autosuspend by default on btusb"
2321
2322 This reverts commit 556ea928f78a390fe16ae584e6433dff304d3014.
2323
2324 Jeff Chua reports that it can cause some bluetooth devices (he mentions
2325 an Bluetooth Intermec scanner) to just stop responding after a while
2326 with messages like
2327
2328 [ 4533.361959] btusb 8-1:1.0: no reset_resume for driver btusb?
2329 [ 4533.361964] btusb 8-1:1.1: no reset_resume for driver btusb?
2330
2331 from the kernel. See also
2332
2333 https://bugzilla.kernel.org/show_bug.cgi?id=26182
2334
2335 for other reports.
2336
2337 Reported-by: Jeff Chua <jeff.chua.linux@gmail.com>
2338 Reported-by: Andrew Meakovski <meako@bigmir.net>
2339 Reported-by: Jim Faulkner <jfaulkne@ccs.neu.edu>
2340 Acked-by: Greg KH <gregkh@suse.de>
2341 Acked-by: Matthew Garrett <mjg@redhat.com>
2342 Acked-by: Gustavo F. Padovan <padovan@profusion.mobi>
2343 Cc: stable@kernel.org (for 2.6.37)
2344 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2345commit e58713724059da7d2982d6ad945192c8fca5b729
2346Author: Linus Torvalds <torvalds@linux-foundation.org>
2347Date: Sun Feb 20 10:03:12 2011 -0800
2348
2349 Revert "tpm_tis: Use timeouts returned from TPM"
2350
2351 This reverts commit 9b29050f8f75916f974a2d231ae5d3cd59792296.
2352
2353 It has caused hibernate regressions, for example Juri Sladby's report:
2354
2355 "I'm unable to hibernate 2.6.37.1 unless I rmmod tpm_tis:
2356 [10974.074587] Suspending console(s) (use no_console_suspend to debug)
2357 [10974.103073] tpm_tis 00:0c: Operation Timed out
2358 [10974.103089] legacy_suspend(): pnp_bus_suspend+0x0/0xa0 returns -62
2359 [10974.103095] PM: Device 00:0c failed to freeze: error -62"
2360
2361 and Rafael points out that some of the new conditionals in that commit
2362 seem to make no sense. This commit needs more work and testing, let's
2363 revert it for now.
2364
2365 Reported-by: Norbert Preining <preining@logic.at>
2366 Reported-and-requested-by: Jiri Slaby <jirislaby@gmail.com>
2367 Cc: Stefan Berger <stefanb@linux.vnet.ibm.com>
2368 Cc: Guillaume Chazarain <guichaz@gmail.com>
2369 Cc: Rajiv Andrade <srajiv@linux.vnet.ibm.com>
2370 Acked-by: Rafael J. Wysocki <rjw@sisk.pl>
2371 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2372commit f00eaeea7a42b5ea327e9ce8839cb0b53d3bdb4e
2373Author: Linus Torvalds <torvalds@linux-foundation.org>
2374Date: Sun Feb 13 07:50:50 2011 -0800
2375
2376 Revert "pci: use security_capable() when checking capablities during config space read"
2377
2378 This reverts commit 47970b1b2aa64464bc0a9543e86361a622ae7c03.
2379
2380 It turns out it breaks several distributions. Looks like the stricter
2381 selinux checks fail due to selinux policies not being set to allow the
2382 access - breaking X, but also lspci.
2383
2384 So while the change was clearly the RightThing(tm) to do in theory, in
2385 practice we have backwards compatibility issues making it not work.
2386
2387 Reported-by: Dave Young <hidave.darkstar@gmail.com>
2388 Acked-by: David Airlie <airlied@linux.ie>
2389 Acked-by: Alex Riesen <raa.lkml@gmail.com>
2390 Cc: Eric Paris <eparis@redhat.com>
2391 Cc: Chris Wright <chrisw@sous-sol.org>
2392 Cc: James Morris <jmorris@namei.org>
2393 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2394commit 83896fb5e51594281720d145164f866ba769abd5
2395Author: Linus Torvalds <torvalds@linux-foundation.org>
2396Date: Mon Jan 17 14:42:34 2011 -0800
2397
2398 Revert "mm: simplify code of swap.c"
2399
2400 This reverts commit d8505dee1a87b8d41b9c4ee1325cd72258226fbc.
2401
2402 Chris Mason ended up chasing down some page allocation errors and pages
2403 stuck waiting on the IO scheduler, and was able to narrow it down to two
2404 commits: commit 744ed1442757 ("mm: batch activate_page() to reduce lock
2405 contention") and d8505dee1a87 ("mm: simplify code of swap.c").
2406
2407 This reverts the second one.
2408
2409 Reported-and-debugged-by: Chris Mason <chris.mason@oracle.com>
2410 Cc: Mel Gorman <mel@csn.ul.ie>
2411 Cc: Andrew Morton <akpm@linux-foundation.org>
2412 Cc: Jens Axboe <jaxboe@fusionio.com>
2413 Cc: linux-mm <linux-mm@kvack.org>
2414 Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
2415 Cc: Andrea Arcangeli <aarcange@redhat.com>
2416 Cc: Shaohua Li <shaohua.li@intel.com>
2417 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2418commit 7a608572a282a74978e10fd6cd63090aebe29f5c
2419Author: Linus Torvalds <torvalds@linux-foundation.org>
2420Date: Mon Jan 17 14:42:19 2011 -0800
2421
2422 Revert "mm: batch activate_page() to reduce lock contention"
2423
2424 This reverts commit 744ed1442757767ffede5008bb13e0805085902e.
2425
2426 Chris Mason ended up chasing down some page allocation errors and pages
2427 stuck waiting on the IO scheduler, and was able to narrow it down to two
2428 commits: commit 744ed1442757 ("mm: batch activate_page() to reduce lock
2429 contention") and d8505dee1a87 ("mm: simplify code of swap.c").
2430
2431 This reverts the first of them.
2432
2433 Reported-and-debugged-by: Chris Mason <chris.mason@oracle.com>
2434 Cc: Mel Gorman <mel@csn.ul.ie>
2435 Cc: Andrew Morton <akpm@linux-foundation.org>
2436 Cc: Jens Axboe <jaxboe@fusionio.com>
2437 Cc: linux-mm <linux-mm@kvack.org>
2438 Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
2439 Cc: Andrea Arcangeli <aarcange@redhat.com>
2440 Cc: Shaohua Li <shaohua.li@intel.com>
2441 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2442commit d8a3515e2a9523f8ed56d1f4537d16338bda2bc2
2443Author: Linus Torvalds <torvalds@linux-foundation.org>
2444Date: Thu Jan 13 17:26:46 2011 -0800
2445
2446 Revert "gpiolib: annotate gpio-intialization with __must_check"
2447
2448 This reverts commit 0fdae42d361bbb431ca0ab0efed5126a94821177, which
2449 wasn't really supposed to go in, and causes lots of annoying warnings.
2450
2451 Quoth Andrew:
2452 "Complete brainfart - I meant to drop that patch ages ago."
2453
2454 Quoth Greg:
2455 "Ick, yeah, that patch isn't ok to go in as-is, all of the callers
2456 need to be fixed up first, which is what I thought we had agreed on..."
2457
2458 Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
2459 Acked-by: Andrew Morton <akpm@linux-foundation.org>
2460 Acked-by: Greg KH <greg@kroah.com>
2461 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2462commit 33e0d57f5d2f079104611be9f3fccc27ef2c6b24
2463Author: Linus Torvalds <torvalds@linux-foundation.org>
2464Date: Fri Nov 19 11:54:40 2010 -0800
2465
2466 Revert "kernel: make /proc/kallsyms mode 400 to reduce ease of attacking"
2467
2468 This reverts commit 59365d136d205cc20fe666ca7f89b1c5001b0d5a.
2469
2470 It turns out that this can break certain existing user land setups.
2471 Quoth Sarah Sharp:
2472
2473 "On Wednesday, I updated my branch to commit 460781b from linus' tree,
2474 and my box would not boot. klogd segfaulted, which stalled the whole
2475 system.
2476
2477 At first I thought it actually hung the box, but it continued booting
2478 after 5 minutes, and I was able to log in. It dropped back to the
2479 text console instead of the graphical bootup display for that period
2480 of time. dmesg surprisingly still works. I've bisected the problem
2481 down to this commit (commit 59365d136d205cc20fe666ca7f89b1c5001b0d5a)
2482
2483 The box is running klogd 1.5.5ubuntu3 (from Jaunty). Yes, I know
2484 that's old. I read the bit in the commit about changing the
2485 permissions of kallsyms after boot, but if I can't boot that doesn't
2486 help."
2487
2488 So let's just keep the old default, and encourage distributions to do
2489 the "chmod -r /proc/kallsyms" in their bootup scripts. This is not
2490 worth a kernel option to change default behavior, since it's so easily
2491 done in user space.
2492
2493 Reported-and-bisected-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
2494 Cc: Marcus Meissner <meissner@suse.de>
2495 Cc: Tejun Heo <tj@kernel.org>
2496 Cc: Eugene Teo <eugeneteo@kernel.org>
2497 Cc: Jesper Juhl <jj@chaosbits.net>
2498 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2499commit 6c2754c28f2388a276fe21edde826f2113c8f60e
2500Author: Linus Torvalds <torvalds@linux-foundation.org>
2501Date: Sat Oct 23 08:14:12 2010 -0700
2502
2503 Revert "tty: Add a new file /proc/tty/consoles"
2504
2505 This reverts commit f4a3e0bceb57466c31757f25e4e0ed108d1299ec. Jiri
2506 Sladby points out that the tty structure we're using may already be
2507 gone, and Al Viro doesn't hold back in complaining about the random
2508 loading of 'filp->private_data' which doesn't have to be a pointer at
2509 all, nor does checking the magic field for TTY_MAGIC prove anything.
2510
2511 Belated review by Al:
2512
2513 "a) global variable depending on stdin of the last opener? Affecting
2514 output of read(2)? Really?
2515
2516 b) iterator is broken; list should be locked in ->start(), unlocked in
2517 ->stop() and *NOT* unlocked/relocked in ->next()
2518
2519 c) ->show() ought to do nothing in case of ->device == NULL, instead
2520 of skipping those in ->next()/->start()
2521
2522 d) regardless of the merits of the bright idea about asterisk at that
2523 line in output *and* regardless of (a), the implementation is not
2524 only atrociously ugly, it's actually very likely to be a roothole.
2525 Verifying that Cthulhu knows what number happens to be address of a
2526 tty_struct by blindly dereferencing memory at that address...
2527 Ouch.
2528
2529 Please revert that crap."
2530
2531 And Christoph pipes in and NAK's the approach of walking fd tables etc
2532 too. So it's pretty unanimous.
2533
2534 Noticed-by: Jri Slaby <jslaby@suse.cz>
2535 Requested-by: Al Viro <viro@zeniv.linux.org.uk>
2536 Cc: Greg Kroah-Hartman <gregkh@suse.de>
2537 Cc: Werner Fink <werner@suse.de>
2538 Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
2539 Cc: Christoph Hellwig <hch@infradead.org>
2540 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2541commit 2069601b3f0ea38170d4b509b89f3ca0a373bdc1
2542Author: Linus Torvalds <torvalds@linux-foundation.org>
2543Date: Thu Aug 12 14:23:04 2010 -0700
2544
2545 Revert "fsnotify: store struct file not struct path"
2546
2547 This reverts commit 3bcf3860a4ff9bbc522820b4b765e65e4deceb3e (and the
2548 accompanying commit c1e5c954020e "vfs/fsnotify: fsnotify_close can delay
2549 the final work in fput" that was a horribly ugly hack to make it work at
2550 all).
2551
2552 The 'struct file' approach not only causes that disgusting hack, it
2553 somehow breaks pulseaudio, probably due to some other subtlety with
2554 f_count handling.
2555
2556 Fix up various conflicts due to later fsnotify work.
2557
2558 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2559commit 8d86dc6a5bcd0f1d5b9364d43843f1bb4b15f57a
2560Author: Linus Torvalds <torvalds@linux-foundation.org>
2561Date: Tue Jun 8 20:16:28 2010 -0700
2562
2563 Revert "drm/i915: Don't enable pipe/plane/VCO early (wait for DPMS on)."
2564
2565 This reverts commit cfecde435dda78248d6fcdc424bed68d5db6be0b, since it
2566 seems to cause some systems to not come up with any video output at all
2567 (or video that only comes on when X starts up).
2568
2569 Fixes bugzilla:
2570
2571 http://bugzilla.kernel.org/show_bug.cgi?id=16163
2572
2573 Reported-and-tested-by: David John <davidjon@xenontk.org>
2574 Tested-by: Nick Bowler <nbowler@elliptictech.com>
2575 Acked-by: Carl Worth <cworth@cworth.org>
2576 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2577commit 386f40c86d6c8d5b717ef20620af1a750d0dacb4
2578Author: Linus Torvalds <torvalds@linux-foundation.org>
2579Date: Sun Jun 6 20:44:04 2010 -0700
2580
2581 Revert "tty: fix a little bug in scrup, vt.c"
2582
2583 This reverts commit 962400e8fd29981a7b166e463dd143b6ac6a3e76, which was
2584 entirely bogus.
2585
2586 The code used to multiply the character offset by "vc->vc_cols", and
2587 that's actually correct, because 'd' itself is an 'unsigned short'. So
2588 the pointer arithmetic already takes the size of a VGA character into
2589 account. Changing it to use vc_size_row (which is just "vc_cols"
2590 shifted up to take the size of the character into account) ends up
2591 multiplying with the VGA character size twice.
2592
2593 This got reported as bugs for various other subsystems, because what it
2594 actually results in is writing the 16-bit vc_video_erase_char pattern
2595 (usually 0x0720: 0x07 is the default attribute, 0x20 is ASCII space)
2596 into some random other allocation.
2597
2598 So Markus ended up reporting this as a ext4 bug, while to Torsten Kaiser
2599 it looked like a problem with KMS or libata. Jeff Chua saw it in
2600 different places.
2601
2602 And finally - Justin Mattock had slab poisoning enabled, and saw it as a
2603 slab poison overwritten. And bisected and reverted this to verify the
2604 buggy commit.
2605
2606 Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de>
2607 Reported-by: Torsten Kaiser <just.for.lkml@googlemail.com>
2608 Reported-by: Jeff Chua <jeff.chua.linux@gmail.com>
2609 Reported-by: Justin P. Mattock <justinmattock@gmail.com>
2610 Reported-bisected-and-tested-by: Justin P. Mattock <justinmattock@gmail.com>
2611 Acked-by: Dave Airlie <airlied@redhat.com>
2612 Cc: Frank Pan <frankpzh@gmail.com>
2613 Cc: Greg Kroah-Hartman <gregkh@suse.de>
2614 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2615commit 35926ff5fba8245bd1c6ac04155048f6f89232b1
2616Author: Linus Torvalds <torvalds@linux-foundation.org>
2617Date: Sun May 30 09:00:03 2010 -0700
2618
2619 Revert "cpusets: randomize node rotor used in cpuset_mem_spread_node()"
2620
2621 This reverts commit 0ac0c0d0f837c499afd02a802f9cf52d3027fa3b, which
2622 caused cross-architecture build problems for all the wrong reasons.
2623 IA64 already added its own version of __node_random(), but the fact is,
2624 there is nothing architectural about the function, and the original
2625 commit was just badly done. Revert it, since no fix is forthcoming.
2626
2627 Requested-by: Stephen Rothwell <sfr@canb.auug.org.au>
2628 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2629commit 13da9e200fe4740b02cd51e07ab454627e228920
2630Author: Linus Torvalds <torvalds@linux-foundation.org>
2631Date: Wed May 26 08:30:15 2010 -0700
2632
2633 Revert "endian: #define __BYTE_ORDER"
2634
2635 This reverts commit b3b77c8caef1750ebeea1054e39e358550ea9f55, which was
2636 also totally broken (see commit 0d2daf5cc858 that reverted the crc32
2637 version of it). As reported by Stephen Rothwell, it causes problems on
2638 big-endian machines:
2639
2640 > In file included from fs/jfs/jfs_types.h:33,
2641 > from fs/jfs/jfs_incore.h:26,
2642 > from fs/jfs/file.c:22:
2643 > fs/jfs/endian24.h:36:101: warning: "__LITTLE_ENDIAN" is not defined
2644
2645 The kernel has never had that crazy "__BYTE_ORDER == __LITTLE_ENDIAN"
2646 model. It's not how we do things, and it isn't how we _should_ do
2647 things. So don't go there.
2648
2649 Requested-by: Stephen Rothwell <sfr@canb.auug.org.au>
2650 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2651commit 218ce7351413b8287a80fab1d7b94906a5559f01
2652Author: Linus Torvalds <torvalds@linux-foundation.org>
2653Date: Tue May 25 16:48:30 2010 -0700
2654
2655 Revert "module: drop the lock while waiting for module to complete initialization."
2656
2657 This reverts commit 480b02df3aa9f07d1c7df0cd8be7a5ca73893455, since
2658 Rafael reports that it causes occasional kernel paging request faults in
2659 load_module().
2660
2661 Dropping the module lock and re-taking it deep in the call-chain is
2662 definitely not the right thing to do. That just turns the mutex from a
2663 lock into a "random non-locking data structure" that doesn't actually
2664 protect what it's supposed to protect.
2665
2666 Requested-and-tested-by: Rafael J. Wysocki <rjw@sisk.pl>
2667 Cc: Rusty Russell <rusty@rustcorp.com.au>
2668 Cc: Brandon Philips <brandon@ifup.org>
2669 Cc: Andrew Morton <akpm@linux-foundation.org>
2670 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2671commit a69eee4988752c7196677958b4ed8f4c2b28499a
2672Author: Linus Torvalds <torvalds@linux-foundation.org>
2673Date: Mon May 24 07:45:43 2010 -0700
2674
2675 Revert "ath9k: Group Key fix for VAPs"
2676
2677 This reverts commit 03ceedea972a82d343fa5c2528b3952fa9e615d5, since it
2678 breaks resume from suspend-to-ram on Rafael's Acer Ferrari One.
2679 NetworkManager thinks everything is ok, but it can't connect to the AP
2680 to get an IP address after the resume.
2681
2682 In fact, it even breaks resume for non-ath9k chipsets: reverting it also
2683 fixes Rafael's Toshiba Protege R500 with the iwlagn driver. As Johannes
2684 says:
2685
2686 "Indeed, this patch needs to be reverted. That mac80211 change is wrong
2687 and completely unnecessary."
2688
2689 Reported-and-requested-by: Rafael J. Wysocki <rjw@sisk.pl>
2690 Acked-by: Johannes Berg <johannes@sipsolutions.net>
2691 Cc: Daniel Yingqiang Ma <yma.cool@gmail.com>
2692 Cc: John W. Linville <linville@tuxdriver.com>
2693 Cc: David Miller <davem@davemloft.net>
2694 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2695commit 769d9968e42c995eaaf61ac5583d998f32e0769a
2696Author: Linus Torvalds <torvalds@linux-foundation.org>
2697Date: Wed May 12 18:39:45 2010 -0700
2698
2699 Revert "PCI: update bridge resources to get more big ranges in PCI assign unssigned"
2700
2701 This reverts commit 977d17bb1749517b353874ccdc9b85abc7a58c2a, because it
2702 can cause problems with some devices not getting any resources at all
2703 when the resource tree is re-allocated.
2704
2705 For an example of this, see
2706
2707 https://bugzilla.kernel.org/show_bug.cgi?id=15960
2708 (originally https://bugtrack.alsa-project.org/alsa-bug/view.php?id=4982)
2709 (lkml thread: http://lkml.org/lkml/2010/4/19/20)
2710
2711 where Peter Henriksson reported his Xonar DX sound card gone, because
2712 the IO port region was no longer allocated.
2713
2714 Reported-bisected-and-tested-by: Peter Henriksson <peter.henriksson@gmail.com>
2715 Requested-by: Andrew Morton <akpm@linux-foundation.org>
2716 Requested-by: Clemens Ladisch <clemens@ladisch.de>
2717 Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org>
2718 Cc: Yinghai Lu <yinghai@kernel.org>
2719 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2720commit 4dc86ae1f925b2121d4e75058675895f83e54c71
2721Author: Linus Torvalds <torvalds@linux-foundation.org>
2722Date: Fri Apr 9 10:05:33 2010 -0700
2723
2724 Revert "memory-hotplug: add 0x prefix to HEX block_size_bytes"
2725
2726 This reverts commit ba168fc37dea145deeb8fa9e7e71c748d2e00d74.
2727
2728 It changes user-visible sysfs interfaces, and breaks some existing user
2729 space applications which apparently rely on the fact that the output
2730 does not contain the "0x" prefix.
2731
2732 Requested-by: Heiko Carstens <heiko.carstens@de.ibm.com>
2733 Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
2734 Acked-by: Wu Fengguang <fengguang.wu@intel.com>
2735 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2736commit b8fa05719ba4349be80ce929237249b57886a203
2737Author: Linus Torvalds <torvalds@linux-foundation.org>
2738Date: Sun Mar 7 09:54:44 2010 -0800
2739
2740 Revert "lib: build list_sort() only if needed"
2741
2742 This reverts commit a069c266ae5fdfbf5b4aecf2c672413aa33b2504.
2743
2744 It turns ou that not only was it missing a case (XFS) that needed it,
2745 but perhaps more importantly, people sometimes want to enable new
2746 modules that they hadn't had enabled before, and if such a module uses
2747 list_sort(), it can't easily be inserted any more.
2748
2749 So rather than add a "select LIST_SORT" to the XFS case, just leave it
2750 compiled in. It's not all _that_ big, after all, and the inconvenience
2751 isn't worth it.
2752
2753 Requested-by: Alexey Dobriyan <adobriyan@gmail.com>
2754 Cc: Christoph Hellwig <hch@infradead.org>
2755 Cc: Don Mullis <don.mullis@gmail.com>
2756 Cc: Andrew Morton <akpm@linux-foundation.org>
2757 Cc: Dave Chinner <david@fromorbit.com>
2758 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2759commit 2f99f5c8f05e02f3df1bb4d93b6704e6f5972872
2760Author: Linus Torvalds <torvalds@linux-foundation.org>
2761Date: Wed Dec 23 15:04:53 2009 -0800
2762
2763 Revert "x86, ucode-amd: Ensure ucode update on suspend/resume after CPU off/online cycle"
2764
2765 This reverts commit 9f15226e75583547aaf542c6be4bdac1060dd425. It's just
2766 wrong, and broke resume for Rafael even on a non-AMD CPU.
2767
2768 As Rafael says:
2769 "... it causes microcode_init_cpu() to be called during resume even for
2770 CPUs for which there's no microcode to apply. That, in turn, results
2771 in executing request_firmware() (on Intel CPUs at least) which doesn't
2772 work at this stage of resume (we have device interrupts disabled, I/O
2773 devices are still suspended and so on).
2774
2775 If I'm not mistaken, the "if (uci->valid)" logic means "if that CPU is
2776 known to us" , so before commit 9f15226e755 microcode_resume_cpu() was
2777 called for all CPUs already in the system during suspend, which was
2778 the right thing to do. The commit changed it so that the CPUs without
2779 microcode to apply are now treated as "unknown", which is not quite
2780 right.
2781
2782 The problem this commit attempted to solve has to be handled
2783 differently."
2784
2785 Bisected-and -requested-by: Rafael J. Wysocki <rjw@sisk.pl>
2786 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2787commit 83f57a11d84460dfe2afdb5a8bc759953428e38b
2788Author: Linus Torvalds <torvalds@linux-foundation.org>
2789Date: Tue Dec 22 14:10:37 2009 -0800
2790
2791 Revert "time: Remove xtime_cache"
2792
2793 This reverts commit 7bc7d637452383d56ba4368d4336b0dde1bb476d, as
2794 requested by John Stultz. Quoting John:
2795
2796 "Petr Titěra reported an issue where he saw odd atime regressions with
2797 2.6.33 where there were a full second worth of nanoseconds in the
2798 nanoseconds field.
2799
2800 He also reviewed the time code and narrowed down the problem: unhandled
2801 overflow of the nanosecond field caused by rounding up the
2802 sub-nanosecond accumulated time.
2803
2804 Details:
2805
2806 * At the end of update_wall_time(), we currently round up the
2807 sub-nanosecond portion of accumulated time when storing it into xtime.
2808 This was added to avoid time inconsistencies caused when the
2809 sub-nanosecond portion was truncated when storing into xtime.
2810 Unfortunately we don't handle the possible second overflow caused by
2811 that rounding.
2812
2813 * Previously the xtime_cache code hid this overflow by normalizing the
2814 xtime value when storing into the xtime_cache.
2815
2816 * We could try to handle the second overflow after the rounding up, but
2817 since this affects the timekeeping's internal state, this would further
2818 complicate the next accumulation cycle, causing small errors in ntp
2819 steering. As much as I'd like to get rid of it, the xtime_cache code is
2820 known to work.
2821
2822 * The correct fix is really to include the sub-nanosecond portion in the
2823 timekeeping accessor function, so we don't need to round up at during
2824 accumulation. This would greatly simplify the accumulation code.
2825 Unfortunately, we can't do this safely until the last three
2826 non-GENERIC_TIME arches (sparc32, arm, cris) are converted (those
2827 patches are in -mm) and we kill off the spots where arches set xtime
2828 directly. This is all 2.6.34 material, so I think reverting the
2829 xtime_cache change is the best approach for now.
2830
2831 Many thanks to Petr for both reporting and finding the issue!"
2832
2833 Reported-by: Petr Titěra <P.Titera@century.cz>
2834 Requested-by: john stultz <johnstul@us.ibm.com>
2835 Cc: Ingo Molnar <mingo@elte.hu>
2836 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2837commit b6e3224fb20954f155e41ec5709b2ab70b50ae2d
2838Author: Linus Torvalds <torvalds@linux-foundation.org>
2839Date: Thu Dec 17 13:23:24 2009 -0800
2840
2841 Revert "task_struct: make journal_info conditional"
2842
2843 This reverts commit e4c570c4cb7a95dbfafa3d016d2739bf3fdfe319, as
2844 requested by Alexey:
2845
2846 "I think I gave a good enough arguments to not merge it.
2847 To iterate:
2848 * patch makes impossible to start using ext3 on EXT3_FS=n kernels
2849 without reboot.
2850 * this is done only for one pointer on task_struct"
2851
2852 None of config options which define task_struct are tristate directly
2853 or effectively."
2854
2855 Requested-by: Alexey Dobriyan <adobriyan@gmail.com>
2856 Acked-by: Andrew Morton <akpm@linux-foundation.org>
2857 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2858commit a2770d86b33024f71df269fde2de096df89d6a48
2859Author: Linus Torvalds <torvalds@linux-foundation.org>
2860Date: Thu Dec 17 12:51:05 2009 -0800
2861
2862 Revert "fix mismerge with Trond's stuff (create_mnt_ns() export is gone now)"
2863
2864 This reverts commit e9496ff46a20a8592fdc7bdaaf41b45eb808d310. Quoth Al:
2865
2866 "it's dependent on a lot of other stuff not currently in mainline
2867 and badly broken with current fs/namespace.c. Sorry, badly
2868 out-of-order cherry-pick from old queue.
2869
2870 PS: there's a large pending series reworking the refcounting and
2871 lifetime rules for vfsmounts that will, among other things, allow to
2872 rip a subtree away _without_ dissolving connections in it, to be
2873 garbage-collected when all active references are gone. It's
2874 considerably saner wrt "is the subtree busy" logics, but it's nowhere
2875 near being ready for merge at the moment; this changeset is one of the
2876 things becoming possible with that sucker, but it certainly shouldn't
2877 have been picked during this cycle. My apologies..."
2878
2879 Noticed-by: Eric Paris <eparis@redhat.com>
2880 Requested-by: Al Viro <viro@ZenIV.linux.org.uk>
2881 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2882commit d4da6c9ccf648f3f1cb5bf9d981a62c253d30e28
2883Author: Linus Torvalds <torvalds@linux-foundation.org>
2884Date: Mon Nov 2 10:15:27 2009 -0800
2885
2886 Revert "ext4: Remove journal_checksum mount option and enable it by default"
2887
2888 This reverts commit d0646f7b636d067d715fab52a2ba9c6f0f46b0d7, as
2889 requested by Eric Sandeen.
2890
2891 It can basically cause an ext4 filesystem to miss recovery (and thus get
2892 mounted with errors) if the journal checksum does not match.
2893
2894 Quoth Eric:
2895
2896 "My hand-wavy hunch about what is happening is that we're finding a
2897 bad checksum on the last partially-written transaction, which is
2898 not surprising, but if we have a wrapped log and we're doing the
2899 initial scan for head/tail, and we abort scanning on that bad
2900 checksum, then we are essentially running an unrecovered filesystem.
2901
2902 But that's hand-wavy and I need to go look at the code.
2903
2904 We lived without journal checksums on by default until now, and at
2905 this point they're doing more harm than good, so we should revert
2906 the default-changing commit until we can fix it and do some good
2907 power-fail testing with the fixes in place."
2908
2909 See
2910
2911 http://bugzilla.kernel.org/show_bug.cgi?id=14354
2912
2913 for all the gory details.
2914
2915 Requested-by: Eric Sandeen <sandeen@redhat.com>
2916 Cc: Theodore Tso <tytso@mit.edu>
2917 Cc: Alexey Fisher <bug-track@fisher-privat.net>
2918 Cc: Maxim Levitsky <maximlevitsky@gmail.com>
2919 Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
2920 Cc: Mathias Burén <mathias.buren@gmail.com>
2921 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2922commit d93a8f829fe1d2f3002f2c6ddb553d12db420412
2923Author: Linus Torvalds <torvalds@linux-foundation.org>
2924Date: Sun Oct 11 15:57:57 2009 -0700
2925
2926 Revert "USB: Work around BIOS bugs by quiescing USB controllers earlier"
2927
2928 This reverts commit db8be50c4307dac2b37305fc59c8dc0f978d09ea, as per
2929
2930 http://bugzilla.kernel.org/show_bug.cgi?id=14374
2931 http://marc.info/?l=linux-kernel&m=125446885705223&w=4
2932
2933 We simply can't do the USB handoff at FIXUP_HEADER time, since it will
2934 often require us to have valid IO mappings etc. But that in turn
2935 requires a whole different approach, not this trivial one-liner.
2936
2937 Maybe we could teach all the USB quirk handoff handlers to only do the
2938 quirk if the device has all its registers set up (since if it isn't
2939 initialized, it's unlikely to be active), but regardless that will need
2940 a whole lot more code than just saying "let's do it really early".
2941
2942 The proper fix is almost certainly to just leave the legacy IOMMU
2943 mappings active until after all devices have been initialized.
2944
2945 Reported-by: Nick Piggin <npiggin@suse.de>
2946 Cc: David Woodhouse <David.Woodhouse@intel.com>
2947 Cc: Greg Kroah-Hartman <gregkh@suse.de>
2948 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2949commit e207e143e2fb6a2790b1ce3687c8aedc3ddc357b
2950Author: Linus Torvalds <torvalds@linux-foundation.org>
2951Date: Wed Sep 30 07:48:37 2009 -0700
2952
2953 Revert "x86, mce: do not compile mcelog message on AMD"
2954
2955 This reverts commit 22223c9b417be5fd0ab2cf9ad17eb7bd1e19f7b9, as
2956 requested by Andi Kleen:
2957
2958 "Obviously kernels compiled with AMD support can still run on non AMD
2959 systems, so messages like this can never be removed at compile time."
2960
2961 Requsted-by: Andi Kleen <andi@firstfloor.org>
2962 Cc: Borislav Petkov <borislav.petkov@amd.com>
2963 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2964commit 4a21b8cb3550f19f838f7c48345fbbf6a0e8536b
2965Author: Linus Torvalds <torvalds@linux-foundation.org>
2966Date: Thu Jul 16 09:14:23 2009 -0700
2967
2968 Revert "ppp: Fix throttling bugs"
2969
2970 This reverts commit a6540f731d506d9e82444cf0020e716613d4c46c, as
2971 requested by Alan:
2972
2973 "... as it was wrong, the pty code is now fixed and the fact this
2974 isn't reverted is breaking pptp setups."
2975
2976 Requested-by: Alan Cox <alan@lxorguk.ukuu.org.uk>
2977 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2978commit 81e4e1ba7ed4a1fdcf0e2ee944f1575010471464
2979Author: Linus Torvalds <torvalds@linux-foundation.org>
2980Date: Sat Jul 11 11:22:34 2009 -0700
2981
2982 Revert "fuse: Fix build error" as unnecessary
2983
2984 This reverts commit 097041e576ee3a50d92dd643ee8ca65bf6a62e21.
2985
2986 Trond had a better fix, which is the parent of this one ("Fix compile
2987 error due to congestion_wait() changes")
2988
2989 Requested-by: Trond Myklebust <Trond.Myklebust@netapp.com>
2990 Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
2991 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2992commit 6c96895e999f5c44a95a5cad6a6e32e7bd6e28b6
2993Author: Linus Torvalds <torvalds@linux-foundation.org>
2994Date: Wed Jul 8 09:20:11 2009 -0700
2995
2996 Revert "fb: Initialize fb_info mutexes in framebuffer_alloc()"
2997
2998 This reverts commit 4148df9b0f38bdd362dd91d52076926c11cbe5a9.
2999
3000 Let's hope that the mm_lock initialization is now correct with all
3001 drivers, following Krzysztof's patches.
3002
3003 Requested-by: Krzysztof Helt <krzysztof.h1@poczta.fm>
3004 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3005commit 236e946b53ffd5e2f5d7e6abebbe72a9f0826d15
3006Author: Linus Torvalds <torvalds@linux-foundation.org>
3007Date: Wed Jun 24 16:23:03 2009 -0700
3008
3009 Revert "PCI: use ACPI _CRS data by default"
3010
3011 This reverts commit 9e9f46c44e487af0a82eb61b624553e2f7118f5b.
3012
3013 Quoting from the commit message:
3014
3015 "At this point, it seems to solve more problems than it causes, so let's
3016 try using it by default. It's an easy revert if it ends up causing
3017 trouble."
3018
3019 And guess what? The _CRS code causes trouble.
3020
3021 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3022commit 752a4787511bf7515f99609ff4ae52341b5bfcde
3023Author: Linus Torvalds <torvalds@linux-foundation.org>
3024Date: Mon Jun 22 11:24:43 2009 -0700
3025
3026 Revert "char: moxa, prevent opening unavailable ports"
3027
3028 This reverts commit a90b037583d5f1ae3e54e9c687c79df82d1d34a4, which
3029 already got fixed as commit f0e8527726b9e56649b9eafde3bc0fbc4dd2dd47:
3030 the same patch (trivial differences) got applied twice.
3031
3032 Requested-by: Alan Cox <alan@lxorguk.ukuu.org.uk>
3033 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3034commit 27704a16c9e0fb4c6b04344c7c4c40ac16148ec0
3035Author: Linus Torvalds <torvalds@linux-foundation.org>
3036Date: Fri Jun 5 10:21:52 2009 -0700
3037
3038 Revert "drm: don't associate _DRM_DRIVER maps with a master"
3039
3040 This reverts commit 6c51d1cfa0a370b48a157163340190cf5fd2346b, which
3041 apparently causes DRI initialization failures on Radeons.
3042
3043 Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de>
3044 Requested-by: Dave Airlie <airlied@gmail.com>
3045 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3046commit 0f181328287db30671e9997329cff71395d4af8b
3047Author: Linus Torvalds <torvalds@linux-foundation.org>
3048Date: Wed May 13 08:29:12 2009 -0700
3049
3050 Revert "Ignore madvise(MADV_WILLNEED) for hugetlbfs-backed regions"
3051
3052 This reverts commit a425a638c858fd10370b573bde81df3ba500e271.
3053
3054 Now that the previous commit removed the "readpage" actor for hugetlb
3055 files, read-ahead will no longer mess up the mapping, and there's no
3056 longer any reason to treat hugetlbfs mappings specially.
3057
3058 Tested-and-acked-by: Mel Gorman <mel@csn.ul.ie>
3059 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3060commit 3d4f16348b77efbf81b7fa186a18a0eb815b6b84
3061Author: Linus Torvalds <torvalds@linux-foundation.org>
3062Date: Mon Apr 27 12:00:27 2009 -0700
3063
3064 Revert "linux.conf.au 2009: Tuz"
3065
3066 This reverts commit 8032b526d1a3bd91ad633dd3a3b5fdbc47ad54f1.
3067
3068 Hey, it was only meant to be a single release. Now they can all die as
3069 far as I'm concerned.
3070
3071 [ Just kidding. They're cute and cuddly.
3072
3073 Except when they have horrible nasty facial diseases. Oh, and I guess
3074 they're not actually that cuddly even when disease-free. ]
3075
3076 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3077commit b0cbc861a3c05e634520b049b5cc27ad6febb51f
3078Author: Linus Torvalds <torvalds@linux-foundation.org>
3079Date: Sat Apr 11 12:45:20 2009 -0700
3080
3081 Revert "ACPI battery: fix async boot oops"
3082
3083 This reverts commit 5d38258ec026921a7b266f4047ebeaa75db358e5, since the
3084 underlying problem got fixed properly in the previous commit ("async:
3085 Fix module loading async-work regression").
3086
3087 Cc: Arkadiusz Miskiewicz <a.miskiewicz@gmail.com>
3088 Cc: Vegard Nossum <vegard.nossum@gmail.com>
3089 Cc: Len Brown <len.brown@intel.com>
3090 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3091commit 2450cf51a1bdba7037e91b1bcc494b01c58aaf66
3092Author: Linus Torvalds <torvalds@linux-foundation.org>
3093Date: Mon Mar 2 16:23:33 2009 -0800
3094
3095 Revert "menu: fix embedded menu snafu"
3096
3097 This reverts commit 155b25bcc28631a5b5230191aa3f56c40dfffa3f, which was
3098 totally wrong - the "embedded" options still exists (very much so) even
3099 on non-embedded platforms.
3100
3101 It's just that we don't bother with actually asking about them when
3102 we're not embedded, we just take their default values (which is usually
3103 'y' - the options add features that may not be worth it in a constrained
3104 environment).
3105
3106 Noticed-by: Andrew Morton <akpm@linux-foundation.org>
3107 Cc: Randy Dunlap <randy.dunlap@oracle.com>
3108 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3109commit 50246dd41ccbcb47beb06d6c1d9355f6b7137a11
3110Author: Linus Torvalds <torvalds@linux-foundation.org>
3111Date: Fri Jan 16 08:14:51 2009 -0800
3112
3113 Revert "PCI PM: Register power state of devices during initialization"
3114
3115 This reverts commit 98e6e286d7b01deb7453b717aa38ebb69d6cefc0, as Yinghai
3116 Lu reports that it breaks kexec with at least the e1000 and e1000e
3117 drivers. The reason is that the shutdown sequence puts the hardware
3118 into D3 sleep, and the commit causes us to claim that it then is in D0
3119 (running) state just because we don't understand the PM capabilities.
3120
3121 Which then later makes "pci_set_power_state()" not do anything, and the
3122 device never wakes up properly and just returns 0xff to everything.
3123
3124 Reported-by: Yinghai Lu <yinghai@kernel.org>
3125 Acked-by: From: Rafael J. Wysocki <rjw@sisk.pl>
3126 Cc: Jesse Barnes <jesse.barnes@intel.com>
3127 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3128commit b5db0e38653bfada34a92f360b4111566ede3842
3129Author: Linus Torvalds <torvalds@linux-foundation.org>
3130Date: Thu Jan 15 15:32:12 2009 -0800
3131
3132 Revert "x86 PAT: remove CPA WARN_ON for zero pte"
3133
3134 This reverts commit 58dab916dfb57328d50deb0aa9b3fc92efa248ff, which
3135 makes my Nehalem come to a nasty crawling almost-halt. It looks like it
3136 turns off caching of regular kernel RAM, with the understandable
3137 slowdown of a few orders of magnitude as a result.
3138
3139 Acked-by: Ingo Molnar <mingo@elte.hu>
3140 Cc: Yinghai Lu <yinghai@kernel.org>
3141 Cc: Peter Anvin <hpa@zytor.com>
3142 Cc: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
3143 Cc: Suresh Siddha <suresh.b.siddha@intel.com>
3144 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3145commit ca7e716c7833aeaeb8fedd6d004c5f5d5e14d325
3146Author: Linus Torvalds <torvalds@linux-foundation.org>
3147Date: Sun Dec 14 15:46:01 2008 -0800
3148
3149 Revert "sched_clock: prevent scd->clock from moving backwards"
3150
3151 This reverts commit 5b7dba4ff834259a5623e03a565748704a8fe449, which
3152 caused a regression in hibernate, reported by and bisected by Fabio
3153 Comolli.
3154
3155 This revert fixes
3156
3157 http://bugzilla.kernel.org/show_bug.cgi?id=12155
3158 http://bugzilla.kernel.org/show_bug.cgi?id=12149
3159
3160 Bisected-by: Fabio Comolli <fabio.comolli@gmail.com>
3161 Requested-by: Rafael J. Wysocki <rjw@sisk.pl>
3162 Acked-by: Dave Kleikamp <shaggy@linux.vnet.ibm.com>
3163 Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
3164 Cc: Ingo Molnar <mingo@elte.hu>
3165 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3166commit 6c34bc2976b30dc8b56392c020e25bae1f363cab
3167Author: Linus Torvalds <torvalds@linux-foundation.org>
3168Date: Wed Dec 10 09:26:17 2008 -0800
3169
3170 Revert "radeonfb: accelerate imageblit and other improvements"
3171
3172 This reverts commit b1ee26bab14886350ba12a5c10cbc0696ac679bf, along with
3173 the "fixes" for it that all just caused problems:
3174
3175 - c4c6fa9891f3d1bcaae4f39fb751d5302965b566 "radeonfb: fix problem with
3176 color expansion & alignment"
3177
3178 - f3179748a157c21d44d929fd3779421ebfbeaa93 "radeonfb: Disable new color
3179 expand acceleration unless explicitely enabled"
3180
3181 because even when disabled, it breaks for people. See
3182
3183 http://bugzilla.kernel.org/show_bug.cgi?id=12191
3184
3185 for the latest example.
3186
3187 Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
3188 Acked-by: David S. Miller <davem@davemloft.net>
3189 Cc: Krzysztof Halasa <khc@pm.waw.pl>
3190 Cc: James Cloos <cloos@jhcloos.com>
3191 Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
3192 Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
3193 Cc: Jean-Luc Coulon <jean.luc.coulon@gmail.com>
3194 Cc: Andrew Morton <akpm@linux-foundation.org>
3195 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3196commit f10a3a32ae2edbe94920ce6827f4c2550eb6ed48
3197Author: Linus Torvalds <torvalds@linux-foundation.org>
3198Date: Fri Dec 5 13:30:03 2008 -0800
3199
3200 Revert "ACPI: battery: Convert discharge energy rate to current properly"
3201
3202 This reverts commit 558073dd56707864f09d563b64e7c37c021e89d2, along with
3203 the failed try to fix the regression it caused ("ACPI: Fix ACPI battery
3204 regression introduced by commit 558073"), which just made things worse.
3205
3206 Commit aaad077638be1a25871bcae5e43952d6b63abfca (that failed "Fix ACPI
3207 battery regression") got the voltage conversion confused, and fixed the
3208 problem with Rafael's battery monitor apparently just by mistake.
3209
3210 So revert them both, getting us back to the 2.6.27 state in this, and
3211 let's revisit it when people understand what's going on.
3212
3213 Noted-by: Paul Martin <pm@debian.org>
3214 Requested-by: Rafael J. Wysocki <rjw@sisk.pl>
3215 Cc: Len Brown <len.brown@intel.com>
3216 Cc: Alexey Starikovskiy <astarikovskiy@suse.de>
3217 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3218commit 33b07db9f38fe73b3895f8d4db8fdee03e3afec3
3219Author: Linus Torvalds <torvalds@linux-foundation.org>
3220Date: Mon Dec 1 07:55:14 2008 -0800
3221
3222 Revert "of_platform_driver noise on sparce"
3223
3224 This reverts commit e669dae6141ff97d3c7566207f5de3b487dcf837, since it
3225 is incomplete, and clashes with fuller patches and the sparc 32/64
3226 unification effort.
3227
3228 Requested-by: David Miller <davem@davemloft.net>
3229 Acked-by: Al Viro <viro@ZenIV.linux.org.uk>
3230 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3231commit 4403b406d4369a275d483ece6ddee0088cc0d592
3232Author: Linus Torvalds <torvalds@linux-foundation.org>
3233Date: Sat Oct 25 19:53:38 2008 -0700
3234
3235 Revert "Call init_workqueues before pre smp initcalls."
3236
3237 This reverts commit a802dd0eb5fc97a50cf1abb1f788a8f6cc5db635 by moving
3238 the call to init_workqueues() back where it belongs - after SMP has been
3239 initialized.
3240
3241 It also moves stop_machine_init() - which needs workqueues - to a later
3242 phase using a core_initcall() instead of early_initcall(). That should
3243 satisfy all ordering requirements, and was apparently the reason why
3244 init_workqueues() was moved to be too early.
3245
3246 Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
3247 Cc: Rusty Russell <rusty@rustcorp.com.au>
3248 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3249commit 93f78da405685a756beeaeae4b5e41fcec39eab3
3250Author: Linus Torvalds <torvalds@linux-foundation.org>
3251Date: Tue Oct 14 12:12:02 2008 -0700
3252
3253 Revert "vt: fix background color on line feed"
3254
3255 This reverts commit c9e587abfdec2c2aaa55fab83bcb4972e2f84f9b, and the
3256 subsequent commits that fixed it up:
3257
3258 - afa9b649 "fbcon: prevent cursor disappearance after switching to 512
3259 character font"
3260
3261 - d850a2fa "vt/fbcon: fix background color on line feed"
3262
3263 - 7fe3915a "vt/fbcon: update scrl_erase_char after 256/512-glyph font
3264 switch"
3265
3266 by request of Alan Cox. Quoth Alan:
3267 "Unfortunately it's wrong and its been causing breakages because
3268 various apps like ncurses expect our previous (and correct)
3269 behaviour."
3270
3271 Alexander sent out a similar patch.
3272
3273 Requested-by: Alan Cox <alan@lxorguk.ukuu.org.uk>
3274 Tested-by: Jan Engelhardt <jengelh@medozas.de>
3275 Cc: Alexander V. Lukyanov <lav@netis.ru>
3276 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3277commit e09e6e2b6a5daf653794926ab50a784b14b6de53
3278Author: Linus Torvalds <torvalds@linux-foundation.org>
3279Date: Tue Oct 7 07:54:34 2008 -0700
3280
3281 Revert "V4L/DVB (8904): cx88: add missing unlock_kernel"
3282
3283 This reverts commit 135aedc38e812b922aa56096f36a3d72ffbcf2fb, as
3284 requested by Hans Verkuil.
3285
3286 It was a patch for 2.6.28 where the BKL was pushed down from v4l core to
3287 the drivers, not for 2.6.27!
3288
3289 Requested-by: Hans Verkuil <hverkuil@xs4all.nl>
3290 Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
3291 Signed-of-by: Linus Torvalds <torvalds@linux-foundation.org>
3292commit e95926d05d028a6bf0ab60b21b484c3d622fdcd1
3293Author: Linus Torvalds <torvalds@linux-foundation.org>
3294Date: Tue Sep 16 19:32:30 2008 -0700
3295
3296 Revert "b43/b43legacy: add RFKILL_STATE_HARD_BLOCKED support"
3297
3298 This reverts commit bc19d6e0b74ef03a3baf035412c95192b54dfc6f, which as
3299 Larry Finger reports causes the radio LED on his system to no longer
3300 respond to rfkill switch events.
3301
3302 Reported-by: Larry Finger <Larry.Finger@lwfinger.net>
3303 Requested-by: John W. Linville <linville@tuxdriver.com>
3304 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3305commit bf7394ccc13fe291d9258f01113b4c61214ddeae
3306Author: Linus Torvalds <torvalds@linux-foundation.org>
3307Date: Fri Sep 5 12:38:09 2008 -0700
3308
3309 Revert "mac80211: Use IWEVASSOCREQIE instead of IWEVCUSTOM"
3310
3311 This reverts commit 087d833e5a9f67ba933cb32eaf5a2279c1a5b47c, which was
3312 reported to break wireless at least in some combinations with 32bit user
3313 space and a 64bit kernel. Alex Williamnson bisected it to this commit.
3314
3315 Reported-and-bisected-by: Alex Williamson <alex.williamson@hp.com>
3316 Acked-by: John W. Linville <linville@tuxdriver.com>
3317 Cc: David Miller <davem@davemloft.net>
3318 Cc: Jouni Malinen <jouni.malinen@atheros.com>
3319 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3320commit b4609472116bb806a95e98d04767189406c74c70
3321Author: Linus Torvalds <torvalds@linux-foundation.org>
3322Date: Fri Aug 29 14:38:03 2008 -0700
3323
3324 Revert "x86: fix HPET regression in 2.6.26 versus 2.6.25, check hpet against BAR, v3"
3325
3326 This reverts commit a2bd7274b47124d2fc4dfdb8c0591f545ba749dd.
3327
3328 It wasn't really right to begin with (there's a better fix for the
3329 problem with e820 reservations clashing with PCI BAR's pending), but it
3330 also actually causes more regressions, so it should be reverted even
3331 before the better fix is finalized.
3332
3333 Rafael reports that this commit broke AHCI detection, and thus causes
3334 the kernel to not boot on his quad core test box.
3335
3336 Reported-and-bisected-by: Rafael J. Wysocki <rjw@sisk.pl>
3337 Cc: Yinghai Lu <yhlu.kernel@gmail.com>
3338 Cc: David Witbrodt <dawitbro@sbcglobal.net>
3339 Cc: Ingo Molnar <mingo@elte.hu>
3340 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3341commit 8560c650f340565b720fd57d1f9c99ab216d99d0
3342Author: Linus Torvalds <torvalds@linux-foundation.org>
3343Date: Wed Aug 27 13:35:31 2008 -0700
3344
3345 Revert "pktcdvd: push BKL down into driver"
3346
3347 This reverts commit 5b6155ee70e9c4d2ad7e6f514c8eee06e2711c3a, because
3348 the block device ioctl's really aren't ready for it.
3349
3350 In particular, the "struct file *" and the "struct inode *" arguments do
3351 not necessarily match, which means that the unlocked version of the
3352 ioctl (that only gets a "struct file *") isn't actually able to handle
3353 the cases it needs to handle.
3354
3355 This fixes bugzilla
3356
3357 http://bugzilla.kernel.org/show_bug.cgi?id=11401
3358
3359 Reported-and-bisected-by: Laurent Riffard <laurent.riffard@free.fr>
3360 Acked-by: Peter Osterlund <petero2@telia.com>
3361 Cc: Alan Cox <alan@redhat.com>
3362 Cc: Andrew Morton <akpm@linux-foundation.org>
3363 Cc: Jens Axboe <jens.axboe@oracle.com>
3364 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3365commit f607e3a03c90e8c050cb0c12ec9967c2925cc812
3366Author: Linus Torvalds <torvalds@linux-foundation.org>
3367Date: Tue Aug 19 13:34:59 2008 -0700
3368
3369 Revert "[CPUFREQ][2/2] preregister support for powernow-k8"
3370
3371 This reverts commit 34ae7f35a21694aa5cb8829dc5142c39d73d6ba0, which has
3372 been reported to cause a number of problems. During suspend and resume,
3373 it apparently causes a crash in a CPU hotplug notifier to happen,
3374 although the exact details are sketchy because of the inability to get
3375 good traces during the suspend sequence.
3376
3377 See buzilla entries
3378
3379 http://bugzilla.kernel.org/show_bug.cgi?id=11296
3380 http://bugzilla.kernel.org/show_bug.cgi?id=11339
3381
3382 for more examples and details.
3383
3384 [ Mark: "Revert the patch for now. I'm still looking into getting a
3385 reliable reproduction and I do not have a fix at this time." ]
3386
3387 Requested-by: Rafael J. Wysocki <rjw@sisk.pl>
3388 Acked-by: Mark Langsdorf <mark.langsdorf@amd.com>
3389 Acked-by: Dave Jones <davej@redhat.com>
3390 Signed-off-by: Linus Torvalds <torvalds@inux-foundation.org>
3391commit 3838f59fc2ea9821f3ea13adb555bfc6ea43c74c
3392Author: Linus Torvalds <torvalds@linux-foundation.org>
3393Date: Mon Aug 11 10:29:11 2008 -0700
3394
3395 Revert "fbcon: bgcolor fix"
3396
3397 This reverts commit 2d04a4a72d7e1519b4838f24bdd4b5d0f3f426dc, which made
3398 it impossible to make the softcursor use the highlight colors.
3399
3400 Yes, the fourth bit should be "blinking", but since we cannot reasonably
3401 blink in fbcon, highlighting it with a bright background is preferable.
3402
3403 Reported-by: Pavel Machek <pavel@suse.cz>
3404 Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
3405 Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
3406 Cc: Antonino A. Daplas <adaplas@pol.net>
3407 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3408commit 685d87f7ccc649ab92b55e18e507a65d0e694eb9
3409Author: Linus Torvalds <torvalds@linux-foundation.org>
3410Date: Wed Aug 6 19:24:47 2008 -0700
3411
3412 Revert "pcm_native.c: remove unused label"
3413
3414 This reverts commit 680db0136e0778a0d7e025af7572c6a8d82279e2. The label
3415 is actually used, but hidden behind CONFIG_SND_DEBUG and the horrible
3416 snd_assert() macro.
3417
3418 That macro could probably be improved to be along the lines of
3419
3420 #define snd_assert(expr, args...) do { if ((void)(expr),0) { args; } } while (0)
3421
3422 or similar to make sure that we always both evaluate 'expr' and parse
3423 'args', but while gcc should optimize it all away, I'm too lazy to
3424 really verify that. So I'll just admit defeat and will continue to live
3425 with the annoying warning.
3426
3427 Noted-by: Robert P. J. Day <rpjday@crashcourse.ca>
3428 Signed-off-by: Linus "Grr.." Torvalds
3429commit d6606683a5e3dac35cb979c7195f54ed827567bd
3430Author: Linus Torvalds <torvalds@linux-foundation.org>
3431Date: Wed Aug 6 12:04:54 2008 -0700
3432
3433 Revert duplicate "mm/hugetlb.c must #include <asm/io.h>"
3434
3435 This reverts commit 7cb93181629c613ee2b8f4ffe3446f8003074842, since we
3436 did that patch twice, and the problem was already fixed earlier by
3437 78a34ae29bf1c9df62a5bd0f0798b6c62a54d520.
3438
3439 Reported-by: Andi Kleen <andi@firstfloor.org>
3440 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3441commit 1a3f7d98e5f50f21ce6fb1406a35531d9596c5c6
3442Author: Linus Torvalds <torvalds@linux-foundation.org>
3443Date: Mon Aug 4 16:50:38 2008 -0700
3444
3445 Revert "UFS: add const to parser token table"
3446
3447 This reverts commit f9247273cb69ba101877e946d2d83044409cc8c5 (and
3448 fb2e405fc1fc8b20d9c78eaa1c7fd5a297efde43 - "fix fs/nfs/nfsroot.c
3449 compilation" - that fixed a missed conversion).
3450
3451 The changes cause problems for at least the sparc build. Let's re-do
3452 them when the exact issues are resolved.
3453
3454 Requested-by: Andrew Morton <akpm@linux-foundation.org>
3455 Requested-by: Steven Whitehouse <swhiteho@redhat.com>
3456 Cc: David Miller <davem@davemloft.net>
3457 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3458commit fca082c9f1e11ec07efa8d2f9f13688521253f36
3459Author: Linus Torvalds <torvalds@linux-foundation.org>
3460Date: Mon Aug 4 16:36:20 2008 -0700
3461
3462 Revert "[SCSI] extend the last_sector_bug flag to cover more sectors"
3463
3464 This reverts commit 2b142900784c6e38c8d39fa57d5f95ef08e735d8, since it
3465 seems to break some other USB storage devices (at least a JMicron USB to
3466 ATA bridge). As such, while it apparently fixes some cardreaders, it
3467 would need to be made conditional on the exact reader it fixes in order
3468 to avoid causing regressions.
3469
3470 Cc: Alan Jenkins <alan-jenkins@tuffmail.co.uk>
3471 Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
3472 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3473commit 09ca8adbe9f724a7e96f512c0039c4c4a1c5dcc0
3474Author: Linus Torvalds <torvalds@linux-foundation.org>
3475Date: Sun Jul 6 10:27:25 2008 -0700
3476
3477 Revert "USB: don't explicitly reenable root-hub status interrupts"
3478
3479 This reverts commit e872154921a6b5256a3c412dd69158ac0b135176.
3480
3481 Andrey Borzenkov reports that it resulted in a totally hung machine for
3482 him when loading the OHCI driver. Extensive netconsole capture with
3483 SysRq output shows that modprobe gets stuck in ohci_hub_status_data()
3484 when probing and enabling the OHCI controller, see for example
3485
3486 http://lkml.org/lkml/2008/7/5/236
3487
3488 for an analysis.
3489
3490 The problem appears to be an interrupt flood triggered by the commit
3491 that gets reverted, and Andrey confirmed that the revert makes things
3492 work for him again.
3493
3494 Reported-and-tested-by: Andrey Borzenkov <arvidjaar@mail.ru>
3495 Acked-by: Alan Stern <stern@rowland.harvard.edu>
3496 Acked-by: David Brownell <david-b@pacbell.net>
3497 Cc: Greg Kroah-Hartman <gregkh@suse.de>
3498 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3499commit 00b41ec2611dc98f87f30753ee00a53db648d662
3500Author: Linus Torvalds <torvalds@linux-foundation.org>
3501Date: Sat May 10 20:43:22 2008 -0700
3502
3503 Revert "semaphore: fix"
3504
3505 This reverts commit bf726eab3711cf192405d21688a4b21e07b6188a, as it has
3506 been reported to cause a regression with processes stuck in __down(),
3507 apparently because some missing wakeup.
3508
3509 Quoth Sven Wegener:
3510 "I'm currently investigating a regression that has showed up with my
3511 last git pull yesterday. Bisecting the commits showed bf726e
3512 "semaphore: fix" to be the culprit, reverting it fixed the issue.
3513
3514 Symptoms: During heavy filesystem usage (e.g. a kernel compile) I get
3515 several compiler processes in uninterruptible sleep, blocking all i/o
3516 on the filesystem. System is an Intel Core 2 Quad running a 64bit
3517 kernel and userspace. Filesystem is xfs on top of lvm. See below for
3518 the output of sysrq-w."
3519
3520 See
3521
3522 http://lkml.org/lkml/2008/5/10/45
3523
3524 for full report.
3525
3526 In the meantime, we can just fix the BKL performance regression by
3527 reverting back to the good old BKL spinlock implementation instead,
3528 since any sleeping lock will generally perform badly, especially if it
3529 tries to be fair.
3530
3531 Reported-by: Sven Wegener <sven.wegener@stealer.net>
3532 Cc: Andrew Morton <akpm@linux-foundation.org>
3533 Cc: Ingo Molnar <mingo@elte.hu>
3534 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3535commit 8d539108560ec121d59eee05160236488266221c
3536Author: Linus Torvalds <torvalds@linux-foundation.org>
3537Date: Thu May 8 18:41:48 2008 -0700
3538
3539 Revert "PCI: remove default PCI expansion ROM memory allocation"
3540
3541 This reverts commit 9f8daccaa05c14e5643bdd4faf5aed9cc8e6f11e, which was
3542 reported to break X startup (xf86-video-ati-6.8.0). See
3543
3544 http://bugs.freedesktop.org/show_bug.cgi?id=15523
3545
3546 for details.
3547
3548 Reported-by: Laurence Withers <l@lwithers.me.uk>
3549 Cc: Gary Hade <garyhade@us.ibm.com>
3550 Cc: Greg KH <greg@kroah.com>
3551 Cc: Jan Beulich <jbeulich@novell.com>
3552 Cc: "Jun'ichi Nomura" <j-nomura@ce.jp.nec.com>
3553 Cc: Andrew Morton <akpm@linux-foundation.org>
3554 Cc: Ingo Molnar <mingo@elte.hu>
3555 Cc: Thomas Gleixner <tglx@linutronix.de>
3556 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3557commit c0a18111e571138747a98af18b3a2124df56a0d1
3558Author: Linus Torvalds <torvalds@linux-foundation.org>
3559Date: Tue May 6 17:09:27 2008 -0700
3560
3561 Revert "uml: fix gcc problem"
3562
3563 This reverts commit 22eecde2f9034764a3fd095eecfa3adfb8ec9a98. Uli
3564 reports that it breaks UML on x86-64 with the Fedora 8 gcc (gcc 4.1.2),
3565 causing a crash on startup. See
3566
3567 http://marc.info/?l=linux-kernel&m=121011722806093&w=2
3568
3569 for a trace.
3570
3571 Reported-by: Ulrich Drepper <drepper@redhat.com>
3572 Cc: Ingo Molnar <mingo@elte.hu>
3573 Cc: Thomas Gleixner <tglx@linutronix.de>
3574 Cc: Andrew Morton <akpm@linux-foundation.org>
3575 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3576commit 099c736a470c8080a166e7a089f1e48e15f9947c
3577Author: Linus Torvalds <torvalds@linux-foundation.org>
3578Date: Mon Apr 7 13:20:08 2008 -0700
3579
3580 Revert "smc91x: fix build breakage from the SMC_GET_MAC_ADDR API upgrade"
3581
3582 This reverts commit 9e6db60825ef7e7999abc610ce256ba768e58162, which was
3583 merged without the API it needed, causing build breakage.
3584
3585 Reported-by: Bryan Wu <cooloney@kernel.org>
3586 Acked-by: Jeff Garzik <jgarzik@redhat.com>
3587 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3588commit 797de7bdb253624c16144f40b72ec65d63cdcca2
3589Author: Linus Torvalds <torvalds@linux-foundation.org>
3590Date: Sat Apr 5 12:14:13 2008 -0700
3591
3592 Revert "ACPI: Ignore _BQC object when registering backlight device"
3593
3594 This reverts commit 7c0ea45be4f114d85ee35caeead8e1660699c46f which
3595 caused a regression with the backlight being set to off when a laptop
3596 doesn't have a _BQC entry to query the actual backlight value. The code
3597 blindly then falls back on a value of 0.
3598
3599 See
3600 http://bugzilla.kernel.org/show_bug.cgi?id=10387
3601 http://lkml.org/lkml/2008/4/2/366
3602
3603 for details.
3604
3605 Bisected-and-reported-by: Andrey Borzenkov <arvidjaar@mail.ru>
3606 Cc: Zhao Yakui <yakui.zhao@intel.com>
3607 Cc: Zhang Rui <rui.zhang@intel.com>
3608 Cc: Len Brown <len.brown@intel.com>
3609 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3610commit e72e9c23ee025a4c063ca112ba0a6059f9ecc9b7
3611Author: Linus Torvalds <torvalds@linux-foundation.org>
3612Date: Thu Mar 27 20:56:33 2008 -0700
3613
3614 Revert "SLUB: remove useless masking of GFP_ZERO"
3615
3616 This reverts commit 3811dbf67162bd08412f1b0e02e554f353e93bdb.
3617
3618 The masking was not at all useless, and it was sensible. We handle
3619 GFP_ZERO in the caller, and passing it down to any page allocator logic
3620 is buggy and wrong.
3621
3622 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3623commit 12c22d6ef299ccf0955e5756eb57d90d7577ac68
3624Author: Linus Torvalds <torvalds@linux-foundation.org>
3625Date: Wed Mar 26 11:22:40 2008 -0700
3626
3627 Revert "PCI: remove transparent bridge sizing"
3628
3629 This reverts commit 8fa5913d54f3b1e09948e6a0db34da887e05ff1f, which
3630 caused various interesting problems for people, including wrong resource
3631 allocations. See for example bugzilla entry "2.6.25-rc2: ohci1394
3632 problem (MMIO broken)" at
3633
3634 http://bugzilla.kernel.org/show_bug.cgi?id=10080
3635
3636 And Gary Hade says:
3637
3638 "The same change had also exposed an issue reported by Paul Martin that
3639 has been causing an Oops while hotplugging ThinkPads to a ThinkPad
3640 Dock II. See
3641
3642 http://lkml.org/lkml/2008/2/19/405
3643 http://bugzilla.kernel.org/show_bug.cgi?id=9961
3644
3645 I have a fix for the ThinkPad docking Oops but if the issue being
3646 discussed here is caused by the transparent bridge sizing removal
3647 change I totally agree that it should be reverted."
3648
3649 The transparent bridge sizing removal change was motivated by
3650 insufficient PCI memory resource for a transparent bridge window that
3651 was being created as a result of expansion ROM(s) being included in
3652 the transparent bridge sizing calculations.
3653
3654 A later "PCI: Remove default PCI expansion ROM memory allocation"
3655 change ( re: http://lkml.org/lkml/2007/12/11/361 ) removes the
3656 expansion ROM(s) from the transparent bridge sizing calculations which
3657 actually resolves the original issue in a different manner. So, even
3658 if the "PCI: remove transparent bridge sizing" is not problematic it
3659 is no longer needed anyway."
3660
3661 Identified-by: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
3662 Tested-by: Thomas Meyer <thomas@m3y3r.de>
3663 Acked-by: Gary Hade <garyhade@us.ibm.com>
3664 Acked-by: Ingo Molnar <mingo@elte.hu>
3665 Cc: Stefan Richter <stefanr@s5r6.in-berlin.de>
3666 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3667commit a345b4ba2086bacc63884e5d72268415a97bcbff
3668Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
3669Date: Mon Mar 3 10:02:44 2008 -0800
3670
3671 Revert "x86: fix pmd_bad and pud_bad to support huge pages"
3672
3673 This reverts commit cded932b75ab0a5f9181ee3da34a0a488d1a14fd.
3674
3675 Arjan bisected down a boot-time hang to this, saying:
3676 ".. it prevents the kernel to finish booting on my (Penryn based)
3677 laptop. The boot stops right after freeing the init memory."
3678
3679 and while it's not clear exactly what triggers it, at this stage we're
3680 better off just reverting it while Ingo tries to figure out what went
3681 wrong.
3682
3683 Requested-by: Arjan van de Ven <arjan@linux.intel.com>
3684 Cc: Hans Rosenfeld <hans.rosenfeld@amd.com>
3685 Cc: Nish Aravamudan <nish.aravamudan@gmail.com>
3686 Acked-by: Ingo Molnar <mingo@elte.hu>
3687 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3688commit 00e962c5408b9f2d0bebd2308673fe982cb9a5fe
3689Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
3690Date: Tue Feb 19 09:08:49 2008 -0800
3691
3692 Revert "SLUB: Alternate fast paths using cmpxchg_local"
3693
3694 This reverts commit 1f84260c8ce3b1ce26d4c1d6dedc2f33a3a29c0c, which is
3695 suspected to be the reason for some very occasional and hard-to-trigger
3696 crashes that usually look related to memory allocation (mostly reported
3697 in networking, but since that's generally the most common source of
3698 shortlived allocations - and allocations in interrupt contexts - that in
3699 itself is not a big clue).
3700
3701 See for example
3702 http://bugzilla.kernel.org/show_bug.cgi?id=9973
3703 http://lkml.org/lkml/2008/2/19/278
3704 etc.
3705
3706 One promising suspicion for what the root cause of bug is (which also
3707 explains why it's so hard to trigger in practice) came from Eric
3708 Dumazet:
3709
3710 "I wonder how SLUB_FASTPATH is supposed to work, since it is affected
3711 by a classical ABA problem of lockless algo.
3712
3713 cmpxchg_local(&c->freelist, object, object[c->offset]) can succeed,
3714 while an interrupt came (on this cpu), and several allocations were
3715 done, and one free was performed at the end of this interruption, so
3716 'object' was recycled.
3717
3718 c->freelist can then contain the previous value (object), but
3719 object[c->offset] was changed by IRQ.
3720
3721 We then put back in freelist an already allocated object."
3722
3723 but another reason for the revert is simply that everybody agrees that
3724 this code was the main suspect just by virtue of the pattern of oopses.
3725
3726 Cc: Torsten Kaiser <just.for.lkml@googlemail.com>
3727 Cc: Christoph Lameter <clameter@sgi.com>
3728 Cc: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
3729 Cc: Pekka Enberg <penberg@cs.helsinki.fi>
3730 Cc: Ingo Molnar <mingo@elte.hu>
3731 Cc: Eric Dumazet <dada1@cosmosbay.com>
3732 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3733commit dbcc2ec60fe15448e5dfe4153f7347ac82b4da73
3734Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
3735Date: Thu Jan 24 13:35:10 2008 -0800
3736
3737 Revert "mac80211: warn when receiving frames with unaligned data"
3738
3739 This reverts commit 81100eb80add328c4d2a377326f15aa0e7236398 for the
3740 release, to avoid the unnecessary warning noise that is only really
3741 relevant to wireless driver developers.
3742
3743 The warning will probably go right back in after I cut the release, but
3744 at least we won't unnecessarily worry users.
3745
3746 Acked-by: John W. Linville <linville@tuxdriver.com>
3747 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3748commit c23f72cae9523d29ff94eec8f30ccbdaf234b20e
3749Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
3750Date: Mon Jan 14 21:21:29 2008 -0800
3751
3752 Revert "writeback: introduce writeback_control.more_io to indicate more io"
3753
3754 This reverts commit 2e6883bdf49abd0e7f0d9b6297fc3be7ebb2250b, as
3755 requested by Fengguang Wu. It's not quite fully baked yet, and while
3756 there are patches around to fix the problems it caused, they should get
3757 more testing. Says Fengguang: "I'll resend them both for -mm later on,
3758 in a more complete patchset".
3759
3760 See
3761
3762 http://bugzilla.kernel.org/show_bug.cgi?id=9738
3763
3764 for some of this discussion.
3765
3766 Requested-by: Fengguang Wu <wfg@mail.ustc.edu.cn>
3767 Cc: Andrew Morton <akpm@linux-foundation.org>
3768 Cc: Peter Zijlstra <peterz@infradead.org>
3769 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3770commit d238998fbfa49f30b02f0a5de5294ca53c58348c
3771Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
3772Date: Tue Jan 8 11:46:37 2008 -0800
3773
3774 Revert "hda_intel suspend latency: shorten codec read"
3775
3776 This reverts commit 57a04513cb35086d54bcb2cb92e6627fc8fa0fae.
3777
3778 Harald Dunkel reports that it broke sound for him:
3779 "Alsa stopped working for me. I still can access /dev/dsp, change the
3780 volume and so on, but the speakers are quiet."
3781
3782 Reverting it fixed things for him.
3783
3784 Reported-and-tested-by: Harald Dunkel <harald.dunkel@t-online.de>
3785 Acked-by: Takashi Iwai <tiwai@suse.de>
3786 Acked-by: Ingo Molnar <mingo@elte.hu>
3787 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3788commit 7b3d9545f9ac8b31528dd2d6d8ec8d19922917b8
3789Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
3790Date: Sun Jan 6 10:17:12 2008 -0800
3791
3792 Revert "scsi: revert "[SCSI] Get rid of scsi_cmnd->done""
3793
3794 This reverts commit ac40532ef0b8649e6f7f83859ea0de1c4ed08a19, which gets
3795 us back the original cleanup of 6f5391c283d7fdcf24bf40786ea79061919d1e1d.
3796
3797 It turns out that the bug that was triggered by that commit was
3798 apparently not actually triggered by that commit at all, and just the
3799 testing conditions had changed enough to make it appear to be due to it.
3800
3801 The real problem seems to have been found by Peter Osterlund:
3802
3803 "pktcdvd sets it [block device size] when opening the /dev/pktcdvd
3804 device, but when the drive is later opened as /dev/scd0, there is
3805 nothing that sets it back. (Btw, 40944 is possible if the disk is a
3806 CDRW that was formatted with "cdrwtool -m 10236".)
3807
3808 The problem is that pktcdvd opens the cd device in non-blocking mode
3809 when pktsetup is run, and doesn't close it again until pktsetup -d is
3810 run. The effect is that if you meanwhile open the cd device,
3811 blkdev.c:do_open() doesn't call bd_set_size() because
3812 bdev->bd_openers is non-zero."
3813
3814 In particular, to repeat the bug (regardless of whether commit
3815 6f5391c283d7fdcf24bf40786ea79061919d1e1d is applied or not):
3816
3817 " 1. Start with an empty drive.
3818 2. pktsetup 0 /dev/scd0
3819 3. Insert a CD containing an isofs filesystem.
3820 4. mount /dev/pktcdvd/0 /mnt/tmp
3821 5. umount /mnt/tmp
3822 6. Press the eject button.
3823 7. Insert a DVD containing a non-writable filesystem.
3824 8. mount /dev/scd0 /mnt/tmp
3825 9. find /mnt/tmp -type f -print0 | xargs -0 sha1sum >/dev/null
3826 10. If the DVD contains data beyond the physical size of a CD, you
3827 get I/O errors in the terminal, and dmesg reports lots of
3828 "attempt to access beyond end of device" errors."
3829
3830 which in turn is because the nested open after the media change won't
3831 cause the size to be set properly (because the original open still holds
3832 the block device, and we only do the bd_set_size() when we don't have
3833 other people holding the device open).
3834
3835 The proper fix for that is probably to just do something like
3836
3837 bdev->bd_inode->i_size = (loff_t)get_capacity(disk)<<9;
3838
3839 in fs/block_dev.c:do_open() even for the cases where we're not the
3840 original opener (but *not* call bd_set_size(), since that will also
3841 change the block size of the device).
3842
3843 Cc: Peter Osterlund <petero2@telia.com>
3844 Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
3845 Cc: Matthew Wilcox <matthew@wil.cx>
3846 Cc: Ingo Molnar <mingo@elte.hu>
3847 Cc: Andrew Morton <akpm@linux-foundation.org>
3848 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3849commit 00684418707c7a1e36ebdedc4b30fbba5d5860b1
3850Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
3851Date: Tue Dec 25 20:16:16 2007 -0800
3852
3853 Revert "x86: fix show cpuinfo cpu number always zero"
3854
3855 This reverts commit fbdcf18df73758b2e187ab94678b30cd5f6ff9f9.
3856
3857 As pointed out by Yanmin Zhang, the problem was already fixed
3858 differently (and correctly), and rather than fix anything, it actually
3859 causes us to create a sub-optimal sched-domains hierarchy (not setting
3860 up the domain belonging to the core) when CONFIG_X86_HT=y.
3861
3862 Requested-by: Yanmin Zhang <yanmin_zhang@linux.intel.com>
3863 Acked-by: Ingo Molnar <mingo@elte.hu>
3864 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3865commit 4af75653031c6d454b4ace47c1536f0d2e727e3e
3866Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
3867Date: Tue Dec 11 19:28:36 2007 -0800
3868
3869 Revert "make bnx2x select ZLIB_INFLATE"
3870
3871 This reverts commit 70eba18b5664f90d7620905e005b89388e5fd94b, as per
3872 Jeff Garzik:
3873
3874 "That was meant for 2.6.25, and actually (due to patching) applied to
3875 a completely unrelated 2.6.24 net driver."
3876
3877 Noted-by: Roland Dreier <rdreier@cisco.com>
3878 Requested-by: Jeff Garzik <jgarzik@pobox.com>
3879 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3880commit af1bff4f1d117f99ad8a88e6532baff475fb44a5
3881Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
3882Date: Mon Dec 10 07:40:54 2007 -0800
3883
3884 Revert "PCI: fix IDE legacy mode resources"
3885
3886 This reverts commit fd6e732186ab522c812ab19c2c5e5befb8ec8115, which
3887 helped up things on MIPS, but was wrong for everything else. As Ralf
3888 Baechle puts it:
3889
3890 "It seems the whole MIPS resource managment is complicated enough (out
3891 of necessity) that only a few people actually grok it. Ioports being
3892 actually memory mapped on MIPS only makes the confusion worse, sigh."
3893
3894 Requested-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
3895 Acked-by: Alan Cox <alan@redhat.com>
3896 Acked-by: Ralf Baechle <ralf@linux-mips.org>
3897 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3898commit 279e1dab949d33737557babfe9f74e0b74fbe39a
3899Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
3900Date: Thu Nov 15 08:44:36 2007 -0800
3901
3902 Revert "skge: fix ram buffer size calculation"
3903
3904 This reverts commit 7fb7ac241162dc51ec0f7644d4a97b2855213c32.
3905
3906 Heikki Orsila reports that it causes a regression:
3907
3908 "Doing
3909
3910 nc host port < /dev/zero
3911
3912 on a sending machine (not skge) to an skge machine that is receiving:
3913
3914 nc -l -p port >/dev/null
3915
3916 with ~60 MiB/s speed, causes the interface go malfunct. A slow
3917 transfer doesn't cause a problem."
3918
3919 See
3920
3921 http://bugzilla.kernel.org/show_bug.cgi?id=9321
3922
3923 for some more information.
3924
3925 There is a workaround (also reported by Heikki):
3926
3927 "After some fiddling, I noticed that not changing the register write
3928 order on patch:
3929
3930 + skge_write32(hw, RB_ADDR(q, RB_END), end);
3931 skge_write32(hw, RB_ADDR(q, RB_WP), start);
3932 skge_write32(hw, RB_ADDR(q, RB_RP), start);
3933 - skge_write32(hw, RB_ADDR(q, RB_END), end);
3934
3935 fixes the visible effect.. Possibly not the root cause of the
3936 problem, but changing the order back fixes networking here."
3937
3938 but that has yet to be ack'ed or tested more widely, so the whole
3939 problem-causing commit gets reverted until this is resolved properly.
3940
3941 Bisected-and-requested-by: Heikki Orsila <shdl@zakalwe.fi>
3942 Cc: Stephen Hemminger <shemminger@linux-foundation.org>
3943 Cc: Jeff Garzik <jeff@garzik.org>
3944 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3945commit 0b832a4b93932103d73c0c3f35ef1153e288327b
3946Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
3947Date: Tue Nov 13 08:07:31 2007 -0800
3948
3949 Revert "ext2/ext3/ext4: add block bitmap validation"
3950
3951 This reverts commit 7c9e69faa28027913ee059c285a5ea8382e24b5d, fixing up
3952 conflicts in fs/ext4/balloc.c manually.
3953
3954 The cost of doing the bitmap validation on each lookup - even when the
3955 bitmap is cached - is absolutely prohibitive. We could, and probably
3956 should, do it only when adding the bitmap to the buffer cache. However,
3957 right now we are better off just reverting it.
3958
3959 Peter Zijlstra measured the cost of this extra validation as a 85%
3960 decrease in cached iozone, and while I had a patch that took it down to
3961 just 17% by not being _quite_ so stupid in the validation, it was still
3962 a big slowdown that could have been avoided by just doing it right.
3963
3964 Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
3965 Cc: Andrew Morton <akpm@linux-foundation.org>
3966 Cc: Aneesh Kumar <aneesh.kumar@linux.vnet.ibm.com>
3967 Cc: Andreas Dilger <adilger@clusterfs.com>
3968 Cc: Mingming Cao <cmm@us.ibm.com>
3969 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3970commit 44048d700bcbfaf4bcca6e2e0a73d89d01ec0878
3971Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
3972Date: Mon Nov 12 14:14:44 2007 -0800
3973
3974 Revert "Bias the placement of kernel pages at lower PFNs"
3975
3976 This reverts commit 5adc5be7cd1bcef6bb64f5255d2a33f20a3cf5be.
3977
3978 Alexey Dobriyan reports that it causes huge slowdowns under some loads,
3979 in his case a "mkfs.ext2" on a 30G partition. With the placement bias,
3980 the mkfs took over four minutes, with it reverted it's back to about ten
3981 seconds for Alexey.
3982
3983 Reported-and-tested-by: Alexey Dobriyan <adobriyan@gmail.com>
3984 Cc: Mel Gorman <mel@csn.ul.ie>
3985 Cc: Andrew Morton <akpm@linux-foundation.org>
3986 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3987commit 3902beb48d369d5e19f66acc2f857865ddc9b3bf
3988Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
3989Date: Sat Nov 10 14:22:46 2007 -0800
3990
3991 Revert "[ARM] 4642/2: netX: default config for netx based boards"
3992
3993 This reverts commit f33bac8dd4573428b94c67149c5607be489092d1, which was
3994 totally bogus.
3995
3996 The arm/configs/netx_defconfig file already existed - in the right
3997 place. Namely under "arch".
3998
3999 Noticed-by: Paul Mundt <lethal@linux-sh.org>
4000 Acked-by: Robert Schwebel <r.schwebel@pengutronix.de>
4001 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4002commit 6a22c57b8d2a62dea7280a6b2ac807a539ef0716
4003Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4004Date: Mon Oct 29 11:36:04 2007 -0700
4005
4006 Revert "x86_64: allocate sparsemem memmap above 4G"
4007
4008 This reverts commit 2e1c49db4c640b35df13889b86b9d62215ade4b6.
4009
4010 First off, testing in Fedora has shown it to cause boot failures,
4011 bisected down by Martin Ebourne, and reported by Dave Jobes. So the
4012 commit will likely be reverted in the 2.6.23 stable kernels.
4013
4014 Secondly, in the 2.6.24 model, x86-64 has now grown support for
4015 SPARSEMEM_VMEMMAP, which disables the relevant code anyway, so while the
4016 bug is not visible any more, it's become invisible due to the code just
4017 being irrelevant and no longer enabled on the only architecture that
4018 this ever affected.
4019
4020 Reported-by: Dave Jones <davej@redhat.com>
4021 Tested-by: Martin Ebourne <fedora@ebourne.me.uk>
4022 Cc: Zou Nan hai <nanhai.zou@intel.com>
4023 Cc: Suresh Siddha <suresh.b.siddha@intel.com>
4024 Cc: Andrew Morton <akpm@linux-foundation.org>
4025 Acked-by: Andy Whitcroft <apw@shadowen.org>
4026 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4027commit e8b8c977734193adedf2b0f607d6252c78e86394
4028Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4029Date: Fri Oct 19 21:25:45 2007 -0700
4030
4031 Revert "kconfig: tristate choices with mixed tristate and boolean values"
4032
4033 This reverts commit a5bf3d891a6a0fb5aa122792d965e3774108b923.
4034
4035 David Brownell notes that this causes a regression visible in the
4036 drivers/usb/gadget Kconfig file:
4037
4038 "That Kconfig hasn't changed (other than adding new drivers), and it's
4039 worked that way for several years now ... so the issue seems to be
4040 changes in menuconfig/kconfig/etc semantics.
4041
4042 The issue is that when USB_GADGET=m, it's no longer possible to
4043 configure peripheral controller drivers as modules ... the
4044 controller drivers can now only be configured for static linkage.
4045
4046 It should be making a choice of one of the controller drivers which
4047 could work on the target system, and allow that driver to be linked
4048 either as a module (ok iff USB_GADGET=m) or statically."
4049
4050 Reverting this commit resolves the problem, and also fixes a second
4051 problem that David noticed: various dependent options couldn't be enabled.
4052
4053 Tested-and-reported-by: David Brownell <david-b@pacbell.net>
4054 Cc: Randy Dunlap <rdunlap@xenotime.net>
4055 Cc: Jan Beulich <jbeulich@novell.com>,
4056 Cc: Andrew Morton <akpm@linux-foundation.org>,
4057 Cc: Sam Ravnborg <sam@ravnborg.org>,
4058 Cc: Roman Zippel <zippel@linux-m68k.org>
4059 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4060commit ff0ce6845bc18292e80ea40d11c3d3a539a3fc5e
4061Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4062Date: Wed Sep 26 15:52:17 2007 -0700
4063
4064 Revert "[PATCH] x86-64: fix x86_64-mm-sched-clock-share"
4065
4066 This reverts commit 184c44d2049c4db7ef6ec65794546954da2c6a0e.
4067
4068 As noted by Dave Jones:
4069 "Linus, please revert the above cset. It doesn't seem to be
4070 necessary (it was added to fix a miscompile in 'make allnoconfig'
4071 which doesn't seem to be repeatable with it reverted) and actively
4072 breaks the ARM SA1100 framebuffer driver."
4073
4074 Requested-by: Dave Jones <davej@redhat.com>
4075 Cc: Russell King <rmk+lkml@arm.linux.org.uk>
4076 Cc: Andrew Morton <akpm@linux-foundation.org>
4077 Cc: Andi Kleen <ak@suse.de>
4078 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4079commit f7f847b01571e86044dc77e03d92f43699652f8d
4080Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4081Date: Wed Sep 26 15:21:33 2007 -0700
4082
4083 Revert "x86-64: Disable local APIC timer use on AMD systems with C1E"
4084
4085 This reverts commit e66485d747505e9d960b864fc6c37f8b2afafaf0, since
4086 Rafael Wysocki noticed that the change only works for his in -mm, not in
4087 mainline (and that both "noapictimer" _and_ "apicmaintimer" are broken
4088 on his hardware, but that's apparently not a regression, just a symptom
4089 of the same issue that causes the automatic apic timer disable to not
4090 work).
4091
4092 It turns out that it really doesn't work correctly on x86-64, since
4093 x86-64 doesn't use the generic clock events for timers yet.
4094
4095 Thanks to Rafal for testing, and here's the ugly details on x86-64 as
4096 per Thomas:
4097
4098 "I just looked into the code and the logic vs. noapictimer on SMP is
4099 completely broken.
4100
4101 On i386 the noapictimer option not only disables the local APIC
4102 timer, it also registers the CPUs for broadcasting via IPI on SMP
4103 systems.
4104
4105 The x86-64 code uses the broadcast only when the local apic timer is
4106 active, i.e. "noapictimer" is not on the command line. This defeats
4107 the whole purpose of "noapictimer". It should be there to make boxen
4108 work, where the local APIC timer actually has a hardware problem,
4109 e.g. the nx6325.
4110
4111 The current implementation of x86_64 only fixes the ACPI c-states
4112 related problem where the APIC timer stops in C3(2), nothing else.
4113
4114 On nx6325 and other AMD X2 equipped systems which have the C1E
4115 enabled we run into the following:
4116
4117 PIT keeps jiffies (and the system) running, but the local APIC timer
4118 interrupts can get out of sync due to this C1E effect.
4119
4120 I don't think this is a critical problem, but it is wrong
4121 nevertheless.
4122
4123 I think it's safe to revert the C1E patch and postpone the fix to the
4124 clock events conversion."
4125
4126 On further reflection, Thomas noted:
4127
4128 "It's even worse than I thought on the first check:
4129
4130 "noapictimer" on the command line of an SMP box prevents _ONLY_ the
4131 boot CPU apic timer from being used. But the secondary CPU is still
4132 unconditionally setting up the APIC timer and uses the non
4133 calibrated variable calibration_result, which is of course 0, to
4134 setup the APIC timer. Wreckage guaranteed."
4135
4136 so we'll just have to wait for the x86 merge to hopefully fix this up
4137 for x86-64.
4138
4139 Tested-and-requested-by: Rafael J. Wysocki <rjw@sisk.pl>
4140 Acked-by: Thomas Gleixner <tglx@linutronix.de>
4141 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4142commit da8f153e51290e7438ba7da66234a864e5d3e1c1
4143Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4144Date: Fri Sep 21 12:09:41 2007 -0700
4145
4146 Revert "x86_64: Quicklist support for x86_64"
4147
4148 This reverts commit 34feb2c83beb3bdf13535a36770f7e50b47ef299.
4149
4150 Suresh Siddha points out that this one breaks the fundamental
4151 requirement that you cannot free page table pages before the TLB caches
4152 are flushed. The quicklists do not give the same kinds of guarantees
4153 that the mmu_gather structure does, at least not in NUMA configurations.
4154
4155 Requested-by: Suresh Siddha <suresh.b.siddha@intel.com>
4156 Acked-by: Andi Kleen <ak@suse.de>
4157 Cc: Andrew Morton <akpm@linux-foundation.org>
4158 Cc: Christoph Lameter <clameter@sgi.com>
4159 Cc: Asit Mallick <asit.k.mallick@intel.com>
4160 Cc: Tony Luck <tony.luck@intel.com>
4161 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4162commit 8eb891fc809b2300137bcd247025628c06c95a63
4163Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4164Date: Mon Aug 20 23:38:44 2007 -0700
4165
4166 Revert "USB: EHCI cpufreq fix"
4167
4168 This reverts commit 196705c9bbc03540429b0f7cf9ee35c2f928a534. It was
4169 reported to cause a regression by Daniel Exner, and Arjan van de Ven
4170 points out that we actually already have infrastructure in place for
4171 setting limits on acceptable DMA latency that would be the much more
4172 correct fix for the problem with some Broadcom EHCI controllers.
4173
4174 Fixed up trivial conflicts due to the changes to support big-endian host
4175 controller descriptors in drivers/usb/host/{ehci-sched.c,ehci.h}.
4176
4177 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4178commit 88ffc3505988196ef5cfdc0278ad89025c2a7b1a
4179Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4180Date: Thu Aug 9 08:10:16 2007 -0700
4181
4182 Revert "genirq: temporary fix for level-triggered IRQ resend"
4183
4184 This reverts commit 0fc4969b866671dfe39b1a9119d0fdc7ea0f63e5. It was
4185 always meant to be temporary, but it's generating more useless noise
4186 than anything else, and we probably should never have done it in the
4187 generic kernel (only had the people involved test it on their own).
4188
4189 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4190commit 602033ed5907a59ce86f709082a35be047743a86
4191Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4192Date: Thu Jul 26 12:07:21 2007 -0700
4193
4194 Revert most of "x86: Fix alternatives and kprobes to remap write-protected kernel text"
4195
4196 This reverts most of commit 19d36ccdc34f5ed444f8a6af0cbfdb6790eb1177.
4197
4198 The way to DEBUG_RODATA interactions with KPROBES and CPU hotplug is to
4199 just not mark the text as being write-protected in the first place.
4200 Both of those facilities depend on rewriting instructions.
4201
4202 Having "helpful" debug facilities that just cause more problem is not
4203 being helpful. It just adds complexity and bugs. Not worth it.
4204
4205 Reported-by: Rafael J. Wysocki <rjw@sisk.pl>
4206 Cc: Andi Kleen <ak@suse.de>
4207 Cc: Andrew Morton <akpm@linux-foundation.org>
4208 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4209commit 2008220879af095d00ca27eb168a55c8595fbc0b
4210Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4211Date: Fri Jul 20 13:28:54 2007 -0700
4212
4213 Revert "sys_time() speedup"
4214
4215 This basically reverts commit 4e44f3497d41db4c3b9051c61410dee8ae4fb49c,
4216 while waiting for it to be re-done more completely. There are cases of
4217 people mixing "time()" with higher-resolution time sources, and we need
4218 to take the nanosecond offsets into account.
4219
4220 Ingo has a patch that does that, but it's still under some discussion.
4221 In the meantime, just revert back to the old simple situation of just
4222 doing the whole exact timesource calculations.
4223
4224 But rather than using do_gettimeofday(), use the internal nanosecond
4225 resolution getnstimeofday(), which at least avoids one unnecessary
4226 conversion (since we really don't care about whether the fractional
4227 seconds are nanoseconds or microseconds - we'll just throw them away).
4228
4229 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4230commit 1985026d32e69ed6dac3ba0ef8ff10366f060ed3
4231Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4232Date: Tue Jul 17 15:57:42 2007 -0700
4233
4234 Revert drivers/ide/ide.c scsi_cmd_ioctl() usage changes
4235
4236 The old IDE driver is not ready to take generic SCSI commands, even if
4237 it uses them for some specific issues (ie the tray open/close ioctls for
4238 IDE CD-ROM's). Pointed out by Bartlomiej.
4239
4240 I'm sure we'll have it fixed properly soon enough, but for now we should
4241 not allow it to cause problems.
4242
4243 Cc: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
4244 Cc: Jens Axboe <jens.axboe@oracle.com>
4245 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4246commit 2e27afb300b56d83bb03fbfa68852b9c1e2920c6
4247Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4248Date: Mon Jul 16 14:31:08 2007 -0700
4249
4250 Revert "[NET]: Fix races in net_rx_action vs netpoll."
4251
4252 This reverts commit 29578624e354f56143d92510fff33a8b2aaa2c03.
4253
4254 Ingo Molnar reports complete breakage with his e1000 card (no
4255 networking, card reports transmit timeouts), and bisected it down to
4256 this commit. Let's figure out what went wrong, but not keep breaking
4257 machines until we do.
4258
4259 Cc: Ingo Molnar <mingo@elte.hu>
4260 Cc: Olaf Kirch <olaf.kirch@oracle.com>
4261 Cc: David Miller <davem@davemloft.net>
4262 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4263commit 8d9107e8c50e1c4ff43c91c8841805833f3ecfb9
4264Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4265Date: Fri Jul 13 16:53:18 2007 -0700
4266
4267 Revert "SELinux: use SECINITSID_NETMSG instead of SECINITSID_UNLABELED for NetLabel"
4268
4269 This reverts commit 9faf65fb6ee2b4e08325ba2d69e5ccf0c46453d0.
4270
4271 It bit people like Michal Piotrowski:
4272
4273 "My system is too secure, I can not login :)"
4274
4275 because it changed how CONFIG_NETLABEL worked, and broke older SElinux
4276 policies.
4277
4278 As a result, quoth James Morris:
4279
4280 "Can you please revert this patch?
4281
4282 We thought it only affected people running MLS, but it will affect others.
4283
4284 Sorry for the hassle."
4285
4286 Cc: James Morris <jmorris@namei.org>
4287 Cc: Stephen Smalley <sds@tycho.nsa.gov>
4288 Cc: Michal Piotrowski <michal.k.k.piotrowski@gmail.com>
4289 Cc: Paul Moore <paul.moore@hp.com>
4290 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4291commit 8b69ad0e690eb5f38c23087247a12e5fde1baeff
4292Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4293Date: Fri Jul 13 10:43:52 2007 -0700
4294
4295 Revert "[CPUFREQ] powernow-k8: clarify number of cores."
4296
4297 This reverts commit 904f7a3f042b5c6aa9e53ce83f2c9de5e33170ff.
4298
4299 As noted by Peter Anvin:
4300
4301 "It causes build failures on i386.
4302
4303 Yet another case of unnecessary divergence between i386 and x86-64
4304 I'm afraid..."
4305
4306 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4307commit 773208946a132fb733ba273ee8562814f828cc28
4308Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4309Date: Thu Jul 12 17:06:50 2007 -0700
4310
4311 Revert "USB: fix gregkh-usb-usb-use-menuconfig-objects"
4312
4313 This reverts commit acb11c8b8020f1f1b2545152020675ef32d09a58.
4314
4315 It was broken. We most certainly *do* want the default to be the old
4316 behaviour (and the common case!), instead of breaking everybodys
4317 configuration and making 99% of all people have to override the default.
4318
4319 What were you guys thinking?
4320
4321 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4322commit de081fa517fed81b0369f2e90ca87c30182879c8
4323Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4324Date: Thu Jul 12 16:40:08 2007 -0700
4325
4326 Revert "[BNX2]: Seems to not need net/tcp.h"
4327
4328 This reverts commit 963bd949b12158d9b5380b718b31c4b33372ed73. The
4329 driver _does_ need the networking header files;
4330
4331 CC [M] drivers/net/bnx2.o
4332 drivers/net/bnx2.c: In function 'bnx2_start_xmit':
4333 drivers/net/bnx2.c:5177: warning: implicit declaration of function 'tcp_optlen'
4334 drivers/net/bnx2.c:5181: error: invalid application of 'sizeof' to incomplete type 'struct ipv6hdr'
4335 drivers/net/bnx2.c:5202: error: invalid application of 'sizeof' to incomplete type 'struct tcphdr'
4336 drivers/net/bnx2.c:5207: warning: implicit declaration of function 'tcp_hdr'
4337 drivers/net/bnx2.c:5207: error: invalid type argument of '->'
4338 make[2]: *** [drivers/net/bnx2.o] Error 1
4339 make[1]: *** [drivers/net] Error 2
4340 make: *** [drivers] Error 2
4341
4342 Cc: Ilpo Jävinen <ilpo.jarvinen@helsinki.fi>
4343 Cc: David Miller <davem@davemloft.net>
4344 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4345commit 1a93fa86bf6eb0d8bed84ef18c8b35be389edee5
4346Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4347Date: Thu May 24 08:50:39 2007 -0700
4348
4349 Revert "HOWTO: bug report addition" (or "HOWTO: mention bughunting")
4350
4351 This reverts commit 722385f75efd82d9f480f0765a1e97a4d83cac0d (or commit
4352 3f27100872b21e4cc70d07b96eeb3611b30bce63, it's your choice ;), since the
4353 same patch to Documentation/HOWTO got added twice because it just kept
4354 applying cleanly.
4355
4356 Noted by Qi Yong.
4357
4358 Cc: Qi Yong <qiyong@fc-cn.com>
4359 Acked-by: Diego Calleja <diegocg@gmail.com>
4360 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4361commit efa5bf1dd2cf3cdee0bfe97cfd76ff2296179ae4
4362Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4363Date: Mon May 21 18:41:17 2007 -0700
4364
4365 Revert "kbuild: make better section mismatch reports on i386, arm and mips"
4366
4367 This reverts commit f892b7d480eec809a5dfbd6e65742b3f3155e50e, which
4368 totally broke the build on x86 with CONFIG_RELOCATABLE (which, as far as
4369 I can tell, is the only case where it should even matter!) due to a
4370 SIGSEGV in modpost.
4371
4372 Cc: Sam Ravnborg <sam@ravnborg.org>
4373 Cc: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
4374 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4375commit b46522394dde7341a5253658096f354b65cbd90d
4376Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4377Date: Thu May 17 20:18:11 2007 -0700
4378
4379 Revert "[PATCH] x86: Drop cc-options call for all options supported in gcc 3.2+"
4380
4381 This reverts commit c8fdd247255a3a027cd9f66dcf93e6847d1d2f85.
4382
4383 It turns out the kernel was correct, and the gcc complaint was a gcc
4384 bug. The preferred stack boundary is expressed not in bytes, but in the
4385 the log2() of the preferred boundary, so "-mpreferred-stack-boundary=2"
4386 is in fact exactly what we want, but a gcc that is compiled for x86-64
4387 will consider it an error (because the 64-bit calling sequence says that
4388 the stack should be 16-byte aligned) even if we are then using "-m32" to
4389 generate 32-bit code.
4390
4391 Noted-by: Mikulas Patocka <mikulas@artax.karlin.mff.cuni.cz>
4392 Cc: Jan Hubicka <jh@suse.cz>
4393 Acked-by: Andi Kleen <ak@suse.de>
4394 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4395commit faa8b6c3c2e1454175609167a25ae525d075f045
4396Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4397Date: Mon May 14 15:24:24 2007 -0700
4398
4399 Revert "ipmi: add new IPMI nmi watchdog handling"
4400
4401 This reverts commit f64da958dfc83335de1d2bef9d3868f30feb4e53.
4402
4403 Andi Kleen is unhappy with the changes, and they really do not seem
4404 worth it. IPMI could use DIE_NMI_IPI instead of the new callback, even
4405 though that ends up having its own set of problems too, mainly because
4406 the IPMI code cannot really know the NMI was from IPMI or not.
4407
4408 Manually fix up conflicts in arch/x86_64/kernel/traps.c and
4409 drivers/char/ipmi/ipmi_watchdog.c.
4410
4411 Cc: Andi Kleen <ak@suse.de>
4412 Cc: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
4413 Cc: Corey Minyard <minyard@acm.org>
4414 Cc: Andrew Morton <akpm@linux-foundation.org>
4415 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4416commit 44ce6294d07555c3d313757105fd44b78208407f
4417Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4418Date: Wed May 9 18:51:36 2007 -0700
4419
4420 Revert "md: improve partition detection in md array"
4421
4422 This reverts commit 5b479c91da90eef605f851508744bfe8269591a0.
4423
4424 Quoth Neil Brown:
4425
4426 "It causes an oops when auto-detecting raid arrays, and it doesn't
4427 seem easy to fix.
4428
4429 The array may not be 'open' when do_md_run is called, so
4430 bdev->bd_disk might be NULL, so bd_set_size can oops.
4431
4432 This whole approach of opening an md device before it has been
4433 assembled just seems to get more and more painful. I think I'm going
4434 to have to come up with something clever to provide both backward
4435 comparability with usage expectation, and sane integration into the
4436 rest of the kernel."
4437
4438 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4439commit 01e73be3c8f254ef19d787f9b6757468175267eb
4440Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4441Date: Tue May 8 20:12:30 2007 -0700
4442
4443 Revert "fbdev: ignore VESA modes if framebuffer is disabled"
4444
4445 This reverts commit 464bdd33e9baad9806c7adbd8dfc37081a55f27e.
4446
4447 Peter Anvin correctly points out that VESA modes have nothing to do with
4448 frame buffers per se - they are often just regular extended text modes.
4449 Disabling them just because we don't have frame buffer support is very
4450 wrong.
4451
4452 Cc: H. Peter Anvin <hpa@zytor.com>
4453 Cc: Antonino A. Daplas <adaplas@gmail.com>,
4454 Cc: Andrew Morton <akpm@linux-foundation.org>
4455 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4456commit e3ebadd95cb621e2c7436f3d3646447ac9d5c16d
4457Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4458Date: Mon May 7 08:44:24 2007 -0700
4459
4460 Revert "[PATCH] x86: __pa and __pa_symbol address space separation"
4461
4462 This was broken. It adds complexity, for no good reason. Rather than
4463 separate __pa() and __pa_symbol(), we should deprecate __pa_symbol(),
4464 and preferably __pa() too - and just use "virt_to_phys()" instead, which
4465 is more readable and has nicer semantics.
4466
4467 However, right now, just undo the separation, and make __pa_symbol() be
4468 the exact same as __pa(). That fixes the bugs this patch introduced,
4469 and we can do the fairly obvious cleanups later.
4470
4471 Do the new __phys_addr() function (which is now the actual workhorse for
4472 the unified __pa()/__pa_symbol()) as a real external function, that way
4473 all the potential issues with compile/link-time optimizations of
4474 constant symbol addresses go away, and we can also, if we choose to, add
4475 more sanity-checking of the argument.
4476
4477 Cc: Eric W. Biederman <ebiederm@xmission.com>
4478 Cc: Vivek Goyal <vgoyal@in.ibm.com>
4479 Cc: Andi Kleen <ak@suse.de>
4480 Cc: Andrew Morton <akpm@linux-foundation.org>
4481 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4482commit 46fcc86dd71d70211e965102fb69414c90381880
4483Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4484Date: Thu Apr 19 18:21:01 2007 -0700
4485
4486 Revert "e1000: fix NAPI performance on 4-port adapters"
4487
4488 This reverts commit 60cba200f11b6f90f35634c5cd608773ae3721b7. It's been
4489 linked to lockups of the e1000 hardware, see for example
4490
4491 https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=229603
4492
4493 but it's likely that the commit itself is not really introducing the
4494 bug, but just allowing an unrelated problem to rear its ugly head (ie
4495 one current working theory is that the code exposes us to a hardware
4496 race condition by decreasing the amount of time we spend in each NAPI
4497 poll cycle).
4498
4499 We'll revert it until root cause is known. Intel has a repeatable
4500 reproduction on two different machines and bus traces of the hardware
4501 doing something bad.
4502
4503 Acked-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
4504 Cc: Jeff Garzik <jeff@garzik.org>
4505 Cc: David S. Miller <davem@davemloft.net>
4506 Cc: Greg KH <gregkh@suse.de>
4507 Cc: Dave Jones <davej@redhat.com>
4508 Cc: Auke Kok <auke-jan.h.kok@intel.com>
4509 Cc: Andrew Morton <akpm@linux-foundation.org>
4510 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4511commit 296d93cd0205433489b0689533426ce0a8cf2dec
4512Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4513Date: Fri Mar 23 08:03:47 2007 -0700
4514
4515 Revert "ACPI: Only use IPI on known broken machines (AMD, Dothan/BaniasPentium M)"
4516
4517 This reverts commit 25496caec111481161e7f06bbfa12a533c43cc6f, which
4518 broke bootup on at least Ingo's ThinkPad T60. Need to figure out
4519 exactly what is wrong before we can re-do the logic.
4520
4521 Requested-by: Ingo Molnar <mingo@elte.hu>
4522 Acked-by: Thomas Gleixner <tglx@linutronix.de>
4523 Cc: Thomas Renninger <trenn@suse.de>
4524 Cc: Len Brown <len.brown@intel.com>
4525 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4526commit be521466feb3bb1cd89de82a2b1d080e9ebd3cb6
4527Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4528Date: Sat Mar 10 14:22:07 2007 -0800
4529
4530 Revert "USB: pxa2xx_udc: fix hardcoded irq number"
4531
4532 This reverts commit d2487cb4257dafb686f682285854fe7f02ca29d8.
4533
4534 Russell King points out that it's obviously bogus, and I have to agree.
4535 Not only does "irq" not even exist in that scope, but we obviously need
4536 to free the irq that we actually requested, and that's IRQ_USB.
4537
4538 Reported-by: Russell King <rmk@arm.linux.org.uk>
4539 Cc: Andrew Morton <akpm@linux-foundation.org>
4540 Cc: Greg KH <greg@kroah.com>
4541 Cc: David Brownell <david-b@pacbell.net>,
4542 Cc: Milan Svoboda <msvoboda@ra.rockwell.com>
4543 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4544commit 38f3323037de22bb0089d08be27be01196e7148b
4545Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4546Date: Tue Mar 6 19:38:01 2007 -0800
4547
4548 Revert "[PATCH] LOG2: Alter get_order() so that it can make use of ilog2() on a constant"
4549
4550 This reverts commit 39d61db0edb34d60b83c5e0d62d0e906578cc707.
4551
4552 The commit was buggy in multiple ways:
4553 - the conversion to ilog2() was incorrect to begin with
4554 - it tested the wrong #defines, so on all architectures but FRV you'd
4555 never see the bug except for constant arguments.
4556 - the new "get_order()" macro used its arguments multiple times, and
4557 didn't even parenthesize them properly
4558 - despite the comments, it was not true that you could use it for
4559 constant initializers, since not all architectures even use the
4560 generic page.h header file.
4561
4562 All of the problems are individually fixable, but it all boils down to:
4563 better just revert it, and re-do it from scratch.
4564
4565 Cc: David Howells <dhowells@redhat.com>
4566 Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
4567 Cc: Paul Mackerras <paulus@samba.org>
4568 Cc: Andrew Morton <akpm@osdl.org>
4569 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4570commit 221dee285ee38099b82437531bcae9fa9cb64cc4
4571Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4572Date: Mon Feb 26 14:55:48 2007 -0800
4573
4574 Revert "[CPUFREQ] constify cpufreq_driver where possible."
4575
4576 This reverts commit aeeddc1435c37fa3fc844f31d39c185b08de4158, which was
4577 half-baked and broken. It just resulted in compile errors, since
4578 cpufreq_register_driver() still changes the 'driver_data' by setting
4579 bits in the flags field. So claiming it is 'const' _really_ doesn't
4580 work.
4581
4582 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4583commit ea3d5226f52ef30f52aa0a04f47f5919c7facacf
4584Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4585Date: Mon Feb 26 09:21:46 2007 -0800
4586
4587 Revert "[PATCH] i386: add idle notifier"
4588
4589 This reverts commit 2ff2d3d74705d34ab71b21f54634fcf50d57bdd5.
4590
4591 Uwe Bugla reports that he cannot mount a floppy drive any more, and Jiri
4592 Slaby bisected it down to this commit.
4593
4594 Benjamin LaHaise also points out that this is a big hot-path, and that
4595 interrupt delivery while idle is very common and should not go through
4596 all these expensive gyrations.
4597
4598 Fix up conflicts in arch/i386/kernel/apic.c and arch/i386/kernel/irq.c
4599 due to other unrelated irq changes.
4600
4601 Cc: Stephane Eranian <eranian@hpl.hp.com>
4602 Cc: Andi Kleen <ak@suse.de>
4603 Cc: Andrew Morton <akpm@osdl.org>
4604 Cc: Uwe Bugla <uwe.bugla@gmx.de>
4605 Cc: Jiri Slaby <jirislaby@gmail.com>
4606 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4607commit b5bf28cde894b3bb3bd25c13a7647020562f9ea0
4608Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4609Date: Wed Feb 21 11:21:44 2007 -0800
4610
4611 Revert "e1000: fix shared interrupt warning message"
4612
4613 This reverts commit d2ed16356ff4fb9de23fbc5e5d582ce580390106.
4614
4615 As Thomas Gleixner reports:
4616 "e1000 is not working anymore. ifup fails permanentely.
4617 ADDRCONF(NETDEV_UP): eth0: link is not ready
4618 nothing else"
4619
4620 The broken commit was identified with "git bisect".
4621
4622 Auke Kok says:
4623 "I think we need to drop this now. The report that says that this
4624 *fixes* something might have been on regular interrupts only. I
4625 currently suspect that it breaks all MSI interrupts, which would make
4626 sense if I look a the code. Very bad indeed."
4627
4628 Cc: Jesse Brandeburg <jesse.brandeburg@intel.com>
4629 Acked-by: Auke Kok <auke-jan.h.kok@intel.com>
4630 Cc: Andrew Morton <akpm@linux-foundation.org>
4631 Cc: Jeff Garzik <jeff@garzik.org>
4632 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4633commit 552ce544edfbe9bce79952a8c0f8d65b7f2d16bb
4634Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4635Date: Tue Feb 13 12:08:18 2007 -0800
4636
4637 Revert "[PATCH] Fix d_path for lazy unmounts"
4638
4639 This reverts commit eb3dfb0cb1f4a44e2d0553f89514ce9f2a9fcaf1.
4640
4641 It causes some strange Gnome problem with dbus-daemon getting stuck, so
4642 we'll revert it until that problem is understood.
4643
4644 Reported by both walt and Greg KH, who both independently git-bisected
4645 the problem to this commit.
4646
4647 Andreas is looking at it.
4648
4649 Reported-by: walt <wa1ter@myrealbox.com>
4650 Reported-by: Greg KH <greg@kroah.com>
4651 Acked-by: Andreas Gruenbacher <agruen@suse.de>
4652 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4653commit 435f8a605d3b56bb96212f4d70b62ecbd0629340
4654Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4655Date: Fri Feb 2 08:07:42 2007 -0800
4656
4657 Revert "[PATCH] fix typo in geode_configre()@cyrix.c"
4658
4659 This reverts commit e4f0ae0ea63caceff37a13f281a72652b7ea71ba.
4660
4661 It's not wrong, but it's not right either, and everybody seems to agree
4662 that the right fix is probably to do the ccr3 write after the ccr4 one
4663 (and that we also should clean it up a bit). And after that we need to
4664 really validate that all the bits that we write to ccr4 actually do
4665 work.
4666
4667 The old 2.6.19 code was insane, and basically didn't change ccr4 at all
4668 (even though it certainly looks like it was the *intent* to do so). So
4669 let's revert the change that may fix things, just because it's not what
4670 was actually ever tested when the code was written, even if it _was_ the
4671 intent.
4672
4673 There's a discussion on http://lkml.org/lkml/2007/1/9/63 that was
4674 started by the patch that now gets reverted, and that discussion may
4675 well contain the proper long-term fix.
4676
4677 Suggested-by: Adrian Bunk <bunk@stusta.de>
4678 Acked-by: Andrew Morton <akpm@linux-foundation.org>
4679 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4680commit 6fd6b17c6d9713f56b5f20903ec3e00fa6cc435e
4681Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4682Date: Wed Jan 31 16:43:36 2007 -0800
4683
4684 Revert "[PATCH] mm: micro optimise zone_watermark_ok"
4685
4686 This reverts commit e80ee884ae0e3794ef2b65a18a767d502ad712ee.
4687
4688 Pawel Sikora had a boot-time oops due to it - because the sign change
4689 invalidates the following comparisons, since 'free_pages' can be
4690 negative.
4691
4692 The micro-optimization just isn't worth it.
4693
4694 Bisected-by: Pawel Sikora <pluto@agmk.net>
4695 Acked-by: Andrew Morton <akpm@osdl.org>
4696 Cc: Nick Piggin <nickpiggin@yahoo.com.au>
4697 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4698commit bcdddfb66cc998252d34758ce4109cedc0d24a5c
4699Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4700Date: Tue Jan 30 14:11:12 2007 -0800
4701
4702 Revert "net: ifb error path loop fix"
4703
4704 This reverts commit 0c0b3ae68ec93b1db5c637d294647d1cca0df763.
4705
4706 Quoth David:
4707
4708 "Jeff, please revert
4709
4710 It's wrong. We had a lengthy analysis of this piece of code
4711 several months ago, and it is correct.
4712
4713 Consider, if we run the loop and we get an error
4714 the following happens:
4715
4716 1) attempt of ifb_init_one(i) fails, therefore we should
4717 not try to "ifb_free_one()" on "i" since it failed
4718 2) the loop iteration first increments "i", then it
4719 check for error
4720
4721 Therefore we must decrement "i" twice before the first
4722 free during the cleanup. One to "undo" the for() loop
4723 increment, and one to "skip" the ifb_init_one() case which
4724 failed."
4725
4726 Reported-by: David Miller <davem@davemloft.net>
4727 Acked-by: Jeff Garzik <jgarzik@pobox.com>
4728 Cc: Andrew Morton <akpm@osdl.org>
4729 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4730commit 444f378b237a0f728f5c4aba752c08d13c209344
4731Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4732Date: Tue Jan 30 13:35:18 2007 -0800
4733
4734 Revert "[PATCH] namespaces: fix exit race by splitting exit"
4735
4736 This reverts commit 7a238fcba0629b6f2edbcd37458bae56fcf36be5 in
4737 preparation for a better and simpler fix proposed by Eric Biederman
4738 (and fixed up by Serge Hallyn)
4739
4740 Acked-by: Serge E. Hallyn <serue@us.ibm.com>
4741 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4742commit 6d3154cc1143f62c3b80d9929caeaec6db8cb451
4743Author: Linus Torvalds <torvalds@woody.linux-foundation.org>
4744Date: Mon Jan 22 08:53:24 2007 -0800
4745
4746 Revert "[PATCH] Fix up mmap_kmem"
4747
4748 This reverts commit 99a10a60ba9bedcf5d70ef81414d3e03816afa3f.
4749
4750 As per Hugh Dickins:
4751
4752 "Nadia Derbey has reported that mmap of /dev/kmem no longer works with
4753 the kernel virtual address as offset, and Franck has confirmed that
4754 his patch came from a misunderstanding of what an offset means to
4755 /dev/kmem - whereas his patch description seems to say that he was
4756 correcting the offset on a few plaforms, there was no such problem to
4757 correct, and his patch was in fact changing its API on all platforms."
4758
4759 Suggested-by: Hugh Dickins <hugh@veritas.com>
4760 Cc: Franck Bui-Huu <fbuihuu@gmail.com>
4761 Cc: Nadia Derbey <Nadia.Derbey@bull.net>
4762 Cc: Andi Kleen <ak@suse.de>
4763 Cc: Arjan van de Ven <arjan@infradead.org>
4764 Cc: Andrew Morton <akpm@osdl.org>
4765 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4766commit fea5f1e19611d94fbf3905875a427c4cb959cd06
4767Author: Linus Torvalds <torvalds@woody.osdl.org>
4768Date: Mon Jan 8 15:04:46 2007 -0800
4769
4770 Revert "[PATCH] x86-64: Try multiple timer variants in check_timer"
4771
4772 This reverts commit b026872601976f666bae77b609dc490d1834bf77, which has
4773 been linked to several problem reports with IO-APIC and the timer.
4774 Machines either don't boot because the timer doesn't happen, or we get
4775 double timer interrupts because we end up double-routing the timer irq
4776 through multiple interfaces.
4777
4778 See for example
4779
4780 http://lkml.org/lkml/2006/12/16/101
4781 http://lkml.org/lkml/2007/1/3/9
4782 http://bugzilla.kernel.org/show_bug.cgi?id=7789
4783
4784 about some of the discussion.
4785
4786 Patches to fix this cleanup exist (and have been confirmed to work fine
4787 at least for some of the affected cases) and we'll revisit it for
4788 2.6.21, but this late in the -rc series we're better off just reverting
4789 the incomplete commit that caused the problems.
4790
4791 Suggested-by: Adrian Bunk <bunk@stusta.de>
4792 Cc: Eric W. Biederman <ebiederm@xmission.com>
4793 Cc: Yinghai Lu <yinghai.lu@amd.com>
4794 Cc: Andrew Morton <akpm@osdl.org>
4795 Cc: Andi Kleen <ak@suse.de>
4796 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
4797commit 90cb28e8f76e57751ffe14abd09c2d53a6aea7c8
4798Author: Linus Torvalds <torvalds@woody.osdl.org>
4799Date: Sat Jan 6 13:28:21 2007 -0800
4800
4801 Revert "[PATCH] binfmt_elf: randomize PIE binaries (2nd try)"
4802
4803 This reverts commit 59287c0913cc9a6c75712a775f6c1c1ef418ef3b.
4804
4805 Hugh Dickins reports that it causes random failures on x86 with SuSE
4806 10.2, and points out
4807
4808 "Isn't that randomization, anywhere from 0x10000 to ELF_ET_DYN_BASE,
4809 sure to place the ET_DYN from time to time just where the comment
4810 says it's trying to avoid? I assume that somehow results in the error
4811 reported."
4812
4813 (where the comment in question is the existing comment in the source
4814 code about mmap/brk clashes).
4815
4816 Suggested-by: Hugh Dickins <hugh@veritas.com>
4817 Acked-by: Marcus Meissner <meissner@suse.de>
4818 Cc: Andrew Morton <akpm@osdl.org>
4819 Cc: Andi Kleen <ak@suse.de>
4820 Cc: Ingo Molnar <mingo@elte.hu>
4821 Cc: Dave Jones <davej@codemonkey.org.uk>
4822 Cc: Arjan van de Ven <arjan@linux.intel.com>
4823 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
4824commit 36f696cd7ffc82c48bf5826bbb4e005370fb9c6a
4825Author: Linus Torvalds <torvalds@woody.osdl.org>
4826Date: Mon Jan 1 10:55:45 2007 -0800
4827
4828 Revert "[PATCH] x86_64: fix boot hang caused by CALGARY_IOMMU_ENABLED_BY_DEFAULT"
4829
4830 This reverts commit a9622f6219ce58faba1417743bf3078501eb3434. Now that
4831 the Calgary code apparently detects itself properly, it's not needed any
4832 more.
4833
4834 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
4835commit b8e6ec865fd1d8838b6ce9516977b65e9f08f876
4836Author: Linus Torvalds <torvalds@woody.osdl.org>
4837Date: Sun Nov 26 16:27:17 2006 -0800
4838
4839 Revert "[PATCH] Enforce "unsigned long flags;" when spinlocking"
4840
4841 This reverts commit ee3ce191e8eaa4cc15c51a28b34143b36404c4f5, since it
4842 broke on at least ARM, MIPS and PA-RISC due to complicated header file
4843 dependencies.
4844
4845 Conflicts in include/linux/spinlock.h (due to the "nested" variety
4846 fixes) fixed up by hand.
4847
4848 Cc: Alexey Dobriyan <adobriyan@gmail.com>
4849 Cc: Ralf Baechle <ralf@linux-mips.org>
4850 Cc: Kyle McMartin <kyle@parisc-linux.org>
4851 Cc: Russell King <rmk+lkml@arm.linux.org.uk>
4852 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
4853commit b976fe19acc565e5137e6f12af7b6633a23e6b7c
4854Author: Linus Torvalds <torvalds@evo.osdl.org>
4855Date: Fri Nov 17 19:31:09 2006 -0800
4856
4857 Revert "ACPI: created a dedicated workqueue for notify() execution"
4858
4859 This reverts commit 37605a6900f6b4d886d995751fcfeef88c4e462c.
4860
4861 Again.
4862
4863 This same bug has now been introduced twice: it was done earlier by
4864 commit b8d35192c55fb055792ff0641408eaaec7c88988, only to be reverted
4865 last time in commit 72945b2b90a5554975b8f72673ab7139d232a121.
4866
4867 We must NOT try to queue up notify handlers to another thread than the
4868 normal ACPI execution thread, because the notifications on some systems
4869 seem to just keep on accumulating until we run out of memory and/or
4870 threads.
4871
4872 Keeping events within the one deferred execution thread automatically
4873 throttles the events properly.
4874
4875 At least the Compaq N620c will lock up completely on the first thermal
4876 event without this patch reverted.
4877
4878 Cc: David Brownell <david-b@pacbell.net>
4879 Cc: Len Brown <len.brown@intel.com>
4880 Cc: Alexey Starikovskiy <alexey.y.starikovskiy@linux.intel.com>
4881 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
4882commit 9a3a04ac386f44175b6a4142eaeab3d4170a57f3
4883Author: Linus Torvalds <torvalds@woody.osdl.org>
4884Date: Tue Nov 14 15:20:51 2006 -0800
4885
4886 Revert "[PATCH] fix Data Acess error in dup_fd"
4887
4888 This reverts commit 0130b0b32ee53dc7add773fcea984f6a26ef1da3.
4889
4890 Sergey Vlasov points out (and Vadim Lobanov concurs) that the bug it was
4891 supposed to fix must be some unrelated memory corruption, and the "fix"
4892 actually causes more problems:
4893
4894 "However, the new code does not look safe in all cases. If some other
4895 task has opened more files while dup_fd() released oldf->file_lock, the
4896 new code will update open_files to the new larger value. But newf was
4897 allocated with the old smaller value of open_files, therefore subsequent
4898 accesses to newf may try to write into unallocated memory."
4899
4900 so revert it.
4901
4902 Cc: Sharyathi Nagesh <sharyath@in.ibm.com>
4903 Cc: Sergey Vlasov <vsu@altlinux.ru>
4904 Cc: Vadim Lobanov <vlobanov@speakeasy.net>
4905 Cc: Andrew Morton <akpm@osdl.org>
4906 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
4907commit de8e7c12430a73654ae3cedbc45428d56c6b777b
4908Author: Linus Torvalds <torvalds@g5.osdl.org>
4909Date: Wed Nov 8 10:09:28 2006 -0800
4910
4911 Revert "[PATCH] i386: Add MMCFG resources to i386 too"
4912
4913 This reverts commit de09bddb9d6f96785be470c832b881e6d72d589f. It tried
4914 to reserve the MMCONFIG mmio memory ranges, but since the MMCONFIG
4915 information is broken and often bogus (which is why we don't dare use it
4916 most of the time _anyway_), it does more harm than good.
4917
4918 Cc: Jeff Chua <jeff.chua.linux@gmail.com>
4919 Cc: Adrian Bunk <bunk@stusta.de>
4920 Cc: Andi Kleen <ak@suse.de>
4921 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
4922commit 80491eb90c750fcd7d13830062f27ae9b7cc5f75
4923Author: Linus Torvalds <torvalds@g5.osdl.org>
4924Date: Sat Nov 4 09:55:00 2006 -0800
4925
4926 Revert unintentional "volatile" changes in ipc/msg.c
4927
4928 Commit 5a06a363ef48444186f18095ae1b932dddbbfa89 ("[PATCH] ipc/msg.c:
4929 clean up coding style") breaks fakeroot on Alpha (variously hangs or
4930 oopses), according to a report by Falk Hueffner.
4931
4932 The fact that the code seems to rely on compiler access ordering through
4933 the use of "volatile" is a pretty certain sign that the code has locking
4934 problems, and we should fix those properly and then remove the whole
4935 "volatile" entirely.
4936
4937 But in the meantime, the movement of "volatile" was unintentional, and
4938 should be reverted.
4939
4940 Cc: Falk Hueffner <falk@debian.org>
4941 Cc: Andrew Morton <akpm@osdl.org>
4942 Acked-by: Ingo Molnar <mingo@elte.hu>
4943 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
4944commit 209ad53bc19667a128d9c51beba873a5c62bff6e
4945Author: Linus Torvalds <torvalds@g5.osdl.org>
4946Date: Sun Oct 29 17:31:49 2006 -0800
4947
4948 Revert "r8169: mac address change support"
4949
4950 This reverts commit a2b98a697fa4e7564f78905b83db122824916cf9.
4951
4952 As per Guennadi Liakhovetski, the mac address change support code breaks
4953 some normal uses (_without_ any address changes), and until it's all
4954 sorted out, we're better off without it.
4955
4956 Says Francois:
4957
4958 "Go revert it.
4959
4960 Despite what I claimed, I can not find a third-party confirmation by
4961 email that it works elsewhere.
4962
4963 It would probably be enough to remove the call to
4964 __rtl8169_set_mac_addr() in rtl8169_hw_start() though."
4965
4966 See also
4967
4968 http://bugzilla.kernel.org/show_bug.cgi?id=6032
4969
4970 Cc: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
4971 Acked-by: Francois Romieu <romieu@fr.zoreil.com>
4972 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
4973commit 0c0e4668e0e65dd1404e8cf066d147235f95561d
4974Author: Linus Torvalds <torvalds@g5.osdl.org>
4975Date: Mon Oct 23 14:25:30 2006 -0700
4976
4977 Revert unintentional and bogus change to drivers/pci/quirks.c
4978
4979 In commit 4e8a5201506423e0241202de1349422af4260296 ("[PKT_SCHED] netem:
4980 Orphan SKB when adding to queue.") Davem mistakenly also included a
4981 temporary diff in his tree that disabled the pci_fixup_video VGA quirk,
4982 which broke sparc64.
4983
4984 This reverts that part of the commit. Sayeth Davem:
4985
4986 "Greg KH has a patch coming to you soon which will move that VGA code
4987 back into x86/x86_64/IA64 specific areas and will fix the sparc64
4988 problem properly."
4989
4990 Special thanks to Claudio Martins <ctpm@ist.utl.pt> for noticing the
4991 error in the first place.
4992
4993 Cc: Claudio Martins <ctpm@ist.utl.pt>
4994 Cc: David Miller <davem@davemloft.net>
4995 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
4996commit b2ef7858db6394b758818358a43c7dd5f232bbcc
4997Author: Linus Torvalds <torvalds@g5.osdl.org>
4998Date: Fri Oct 20 10:40:48 2006 -0700
4999
5000 Revert "[mv643xx] Add pci device table for auto module loading."
5001
5002 This reverts commit 4596c75c23dde2623cbeec69357d5eb13d28387e as
5003 requested by Olaf Hering. It causes compile errors, and says Olaf:
5004
5005 "This change is also wrong, the autoloading works perfect with 2.6.18,
5006 no need to add random PCI ids.
5007
5008 See commit a0245f7ad5214cb00131d7cd176446e067c913dc, platform devices
5009 have now a modalias entry in sysfs. The network card is not a PCI
5010 device."
5011
5012 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5013commit e24bb60e11e3fe9858b71874a4ac59333adbc4fc
5014Author: Linus Torvalds <torvalds@g5.osdl.org>
5015Date: Sun Oct 8 15:38:22 2006 -0700
5016
5017 Revert "[POWERPC] Don't get PCI IRQ from OF for devices with no IRQ"
5018
5019 This reverts commit 41550c5128150175197257b6ceab2cd50dea7b51.
5020
5021 Quoth Ben Herrenschmidt:
5022 "Please revert this one for now. It seems to break G5s :( Looks like
5023 PCI cells inside Apple IO ASICs don't have a PCI_INTERRUPT_LINE set.
5024 I need to figure out a better fix."
5025
5026 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5027commit 79e453d49bd49ba1b576f89310cc565c9e4ca379
5028Author: Linus Torvalds <torvalds@g5.osdl.org>
5029Date: Tue Sep 19 08:15:22 2006 -0700
5030
5031 Revert mmiocfg heuristics and blacklist changes
5032
5033 This reverts commits 11012d419cfc0e0f78ca356aca03674217910124 and
5034 40dd2d20f220eda1cd0da8ea3f0f9db8971ba237, which allowed us to use the
5035 MMIO accesses for PCI config cycles even without the area being marked
5036 reserved in the e820 memory tables.
5037
5038 Those changes were needed for EFI-environment Intel macs, but broke some
5039 newer Intel 965 boards, so for now it's better to revert to our old
5040 2.6.17 behaviour and at least avoid introducing any new breakage.
5041
5042 Andi Kleen has a set of patches that work with both EFI and the broken
5043 Intel 965 boards, which will be applied once they get wider testing.
5044
5045 Cc: Arjan van de Ven <arjan@infradead.org>
5046 Cc: Edgar Hucek <hostmaster@ed-soft.at>
5047 Cc: Andi Kleen <ak@suse.de>
5048 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5049commit 2f1b9250516aa8744496abbbbb8a6066cb2b56db
5050Author: Linus Torvalds <torvalds@evo.osdl.org>
5051Date: Wed Jul 12 21:04:16 2006 -0700
5052
5053 Revert "[PATCH] pcmcia: Make ide_cs work with the memory space of CF-Cards if IO space is not available"
5054
5055 This reverts commit 5040cb8b7e61b7a03e8837920b9eb2c839bb1947.
5056
5057 It breaks previously working ide-cs PIO configurations, causing problems
5058 like
5059
5060 ide2: I/O resource 0xF883200E-0xF883200E not free.
5061 ide2: ports already in use, skipping probe
5062
5063 rather than a working kernel.
5064
5065 Cc: Thomas Kleffel <tk@maintech.de>
5066 Cc: Dominik Brodowski <linux@dominikbrodowski.net>
5067 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5068commit 953969ddf5b049361ed1e8471cc43dc4134d2a6f
5069Author: Linus Torvalds <torvalds@g5.osdl.org>
5070Date: Sun Jul 9 08:47:46 2006 -0700
5071
5072 Revert "ACPI: dock driver"
5073
5074 This reverts commit a5e1b94008f2a96abf4a0c0371a55a56b320c13e.
5075
5076 Adrian Bunk points out that it has build errors, and apparently no
5077 maintenance. Throw it out.
5078
5079 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5080commit d38b69689c349f35502b92e20dafb30c62d49d63
5081Author: Linus Torvalds <torvalds@g5.osdl.org>
5082Date: Mon Jun 26 16:59:26 2006 -0700
5083
5084 Revert "kbuild: fix make -rR breakage"
5085
5086 This reverts commit e5c44fd88c146755da6941d047de4d97651404a9.
5087
5088 Thanks to Daniel Ritz and Michal Piotrowski for noticing the problem.
5089
5090 Daniel says:
5091
5092 "[The] reason is a recent change that made modules always shows as
5093 module.mod. it breaks modprobe and probably many scripts..besides
5094 lsmod looking horrible
5095
5096 stuff like this in modprobe.conf:
5097 install pcmcia_core /sbin/modprobe --ignore-install pcmcia_core; /sbin/modprobe pcmcia
5098 makes modprobe fork/exec endlessly calling itself...until oom
5099 interrupts it"
5100
5101 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5102commit 09c0dc68625c06f5b1e786aad0d5369b592179e6
5103Author: Linus Torvalds <torvalds@g5.osdl.org>
5104Date: Mon Jun 26 11:55:42 2006 -0700
5105
5106 Revert "[PATCH] kthread: update loop.c to use kthread"
5107
5108 This reverts commit c7b2eff059fcc2d1b7085ee3d84b79fd657a537b.
5109
5110 Hugh Dickins explains:
5111
5112 "It seems too little tested: "losetup -d /dev/loop0" fails with
5113 EINVAL because nothing sets lo_thread; but even when you patch
5114 loop_thread() to set lo->lo_thread = current, it can't survive
5115 more than a few dozen iterations of the loop below (with a tmpfs
5116 mounted on /tst):
5117
5118 j=0
5119 cp /dev/zero /tst
5120 while :
5121 do
5122 let j=j+1
5123 echo "Doing pass $j"
5124 losetup /dev/loop0 /tst/zero
5125 mkfs -t ext2 -b 1024 /dev/loop0 >/dev/null 2>&1
5126 mount -t ext2 /dev/loop0 /mnt
5127 umount /mnt
5128 losetup -d /dev/loop0
5129 done
5130
5131 it collapses with failed ioctl then BUG_ON(!bio).
5132
5133 I think the original lo_done completion was more subtle and safe
5134 than the kthread conversion has allowed for."
5135
5136 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5137commit 3448097fccdce4ea8f0fcad4f37f502a8cd72e68
5138Author: Linus Torvalds <torvalds@g5.osdl.org>
5139Date: Sun Jun 25 18:41:00 2006 -0700
5140
5141 Revert "swsusp special saveable pages support" commits
5142
5143 This reverts commits
5144
5145 3e3318dee0878d42ed62a19c292a2ac284135db3 [PATCH] swsusp: x86_64 mark special saveable/unsaveable pages
5146 b6370d96e09944c6e3ae8d5743ca8a8ab1f79f6c [PATCH] swsusp: i386 mark special saveable/unsaveable pages
5147 ce4ab0012b32c1a4a1d6e934aeb73bf3151c48d9 [PATCH] swsusp: add architecture special saveable pages support
5148
5149 because not only do they apparently cause page faults on x86, the
5150 infrastructure doesn't compile on powerpc.
5151
5152 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5153commit 83626b01275d0228516b4d97da008328fc37c934
5154Author: Linus Torvalds <torvalds@g5.osdl.org>
5155Date: Sat Jun 24 17:47:09 2006 -0700
5156
5157 Revert "[PATCH] usb: drivers/usb/core/devio.c dereferences a userspace pointer"
5158
5159 This reverts commit 786dc1d3d7333f269e17d742886eac2188a2d9cc.
5160
5161 As Al so eloquently points out, the patch is crap. The old code was fine,
5162 the new code was bogus.
5163
5164 It never dereferenced a user pointer, the "->" operator was to an array
5165 member, which gives the _address_ of the member (in user space), not an
5166 actual dereference at all.
5167
5168 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5169commit 9a8fca0499c611ab37b5c0d4481ca09d3f6e8101
5170Author: Linus Torvalds <torvalds@g5.osdl.org>
5171Date: Tue May 30 20:32:15 2006 -0700
5172
5173 Revert "[PATCH] i386/x86_64: Force pci=noacpi on HP XW9300"
5174
5175 This reverts commit 5491d0f3e206beb95eeb506510d62a1dab462df1.
5176
5177 As per Andi:
5178
5179 "After some discussion with people who have the affected system it
5180 seems best to revert for 2.6.17. It broke a common BIOS workaround
5181 and PCI-X still doesn't work. Alternative is for people to change
5182 the BIOS which seems to be better right now."
5183
5184 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5185commit f1adad78dd2fc8edaa513e0bde92b4c64340245c
5186Author: Linus Torvalds <torvalds@g5.osdl.org>
5187Date: Sun May 21 18:54:09 2006 -0700
5188
5189 Revert "[PATCH] sched: fix interactive task starvation"
5190
5191 This reverts commit 5ce74abe788a26698876e66b9c9ce7e7acc25413 (and its
5192 dependent commit 8a5bc075b8d8cf7a87b3f08fad2fba0f5d13295e), because of
5193 audio underruns.
5194
5195 Reported by Rene Herman <rene.herman@keyaccess.nl>, who also pinpointed
5196 the exact cause of the underruns:
5197
5198 "Audio underruns galore, with only ogg123 and firefox (browsing the
5199 GIT tree online is also a nice trigger by the way).
5200
5201 If I back it out, everything is fine for me again."
5202
5203 Cc: Rene Herman <rene.herman@keyaccess.nl>
5204 Cc: Mike Galbraith <efault@gmx.de>
5205 Acked-by: Con Kolivas <kernel@kolivas.org>
5206 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5207commit 9be2f7c38e0bd64e8a0f74ea68df1e73e2ddfcc3
5208Author: Linus Torvalds <torvalds@g5.osdl.org>
5209Date: Sat May 13 08:01:23 2006 -0700
5210
5211 Revert "[PATCH] i386: export: memory more than 4G through /proc/iomem"
5212
5213 This reverts commit 10dbe196a8da6b3196881269c6639c0ec11c36cb.
5214
5215 The resource struct is still 32-bit, so trying to save a 64-bit memory
5216 size there obviously won't work.
5217
5218 When we merge the 64-bit resource series, we can re-enable this.
5219
5220 Thanks to Sachin Sant and Maneesh Soni for debugging
5221
5222 Cc: Maneesh Soni <maneesh@in.ibm.com>
5223 Cc: Sachin Sant <sachinp@in.ibm.com>
5224 Cc: Russell King <rmk+lkml@arm.linux.org.uk>
5225 Cc: Sharyathi Nagesh <sharyath@in.ibm.com>
5226 Cc: Arjan van de Ven <arjan@infradead.org>
5227 Cc: Vivek Goyal <vgoyal@in.ibm.com>
5228 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5229commit 1a2acc9e9214699a99389e323e6686e9e0e2ca67
5230Author: Linus Torvalds <torvalds@g5.osdl.org>
5231Date: Fri May 12 12:08:46 2006 -0700
5232
5233 Revert "[BLOCK] Fix oops on removal of SD/MMC card"
5234
5235 This reverts commit 56cf6504fc1c0c221b82cebc16a444b684140fb7.
5236
5237 Both Erik Mouw and Andrew Vasquez independently pinpointed this commit
5238 as causing problems, where the slab cache for a driver is never released
5239 (most obviously causing problems when immediately re-loading that
5240 driver, resulting in a "kmem_cache_create: duplicate cache <xyz>"
5241 message, but it can also cause other trouble).
5242
5243 James Bottomley dug into it, and reports:
5244
5245 "OK, here's the scoop. The problem patch adds a get of driverfs_dev in
5246 add_disk(), but doesn't put it again until disk_release() (which occurs
5247 on final put_disk() of the gendisk).
5248
5249 However, in SCSI, the driverfs_dev is the sdev_gendev. That means
5250 there's a reference held on sdev_gendev until final disk put.
5251 Unfortunately, we use the driver model driver_remove to trigger
5252 del_gendisk (which removes the gendisk from visibility and decrements
5253 the refcount), so we've introduced an unbreakable deadlock in the
5254 reference counting with this.
5255
5256 I suggest simply reversing this patch at the moment. If Russell and
5257 Jens can tell me what they're trying to do I'll see if there's another
5258 way to do it."
5259
5260 so hereby the patch gets reverted, waiting for a better fix.
5261
5262 Cc: Jens Axboe <axboe@suse.de>
5263 Cc: Russell King <rmk@arm.linux.org.uk>
5264 Cc: James Bottomley <James.Bottomley@SteelEye.com>
5265 Cc: Erik Mouw <erik@harddisk-recovery.com>
5266 Cc: Andrew Vasquez <andrew.vasquez@qlogic.com>
5267 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5268commit 601e7f024edbea8018de34c83a7398623214e636
5269Author: Linus Torvalds <torvalds@g5.osdl.org>
5270Date: Mon May 8 13:38:42 2006 -0700
5271
5272 Revert "kbuild: fix modpost segfault for 64bit mipsel kernel"
5273
5274 This reverts commit c8d8b837ebe4b4f11e1b0c4a2bdc358c697692ed, which
5275 caused problems for the x86 build. Quoth Sam:
5276
5277 "It was discussed on mips list but apparently the fix was bogus. I
5278 will not have time to look into it so mips can carry this local fix
5279 until we get a proper fix in mainline."
5280
5281 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5282commit e0f4ab8a1741193891f096aa63df9ac8672af54c
5283Author: Linus Torvalds <torvalds@g5.osdl.org>
5284Date: Tue Mar 21 14:51:37 2006 -0800
5285
5286 Revert "V4L/DVB (3543): Fix Makefile to adapt to bt8xx/ conversion"
5287
5288 This reverts commit 08f1d0b99f4e2203935d86640a7fec5c233b777c
5289
5290 The "bt8xx/ conversion" for drivers/video/ hasn't actually percolated
5291 all the way to this tree, so the Makefile change escaped too soon.
5292
5293 Build breakage noticed by Jeff Garzik <jeff@garzik.org>
5294
5295 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5296commit cbf0ec6ee001ab89471397ac84105b2e3b2fd986
5297Author: Linus Torvalds <torvalds@g5.osdl.org>
5298Date: Tue Mar 14 08:01:47 2006 -0800
5299
5300 Revert "[PATCH] x86-64: Fix up handling of non canonical user RIPs"
5301
5302 This reverts commit c33d4568aca9028a22857f94f5e0850012b6444b.
5303
5304 Andrew Clayton and Hugh Dickins report that it's broken for them and
5305 causes strange page table and slab corruption, and spontaneous reboots.
5306
5307 Let's get it right next time.
5308
5309 Cc: Andrew Clayton <andrew@rootshell.co.uk>
5310 Cc: Hugh Dickins <hugh@veritas.com>
5311 Cc: Andi Kleen <ak@suse.de>
5312 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5313commit 637029c6cb5efcbaa3d5831af4c1972bdd629779
5314Author: Linus Torvalds <torvalds@g5.osdl.org>
5315Date: Mon Feb 27 20:41:56 2006 -0800
5316
5317 Revert "[PATCH] x86_64: Only do the clustered systems have unsynchronized TSC assumption on IBM systems"
5318
5319 This reverts commit 13a229abc25640813f1480c0478dfc6bdbc1c19e.
5320
5321 Quoth Andi:
5322 "After some consideration and feedback from various people it turns
5323 out this wasn't that good an idea. It has some problems and needs
5324 more work. Since it was only an optimization anyways it's best to
5325 just back it out again for now."
5326
5327 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5328commit 15508d22d00277a1f2a1022dce38f2772c810d32
5329Author: Linus Torvalds <torvalds@g5.osdl.org>
5330Date: Thu Feb 9 08:00:14 2006 -0800
5331
5332 Revert "[PATCH] kconfig: detect if -lintl is needed when linking conf,mconf"
5333
5334 This reverts commit 5e375bc7d586e0df971734a5a5f1f080ffd89b68.
5335
5336 Kyle McMartin steps on his soap-box:
5337
5338 "Sigh. Can everyone please stop assuming gcc can output to /dev/null?
5339 On several platforms, ld tries to lseek in the output file, and fails
5340 if it can't."
5341
5342 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5343commit cef5076987dd545ac74f4efcf1c962be8eac34b0
5344Author: Linus Torvalds <torvalds@g5.osdl.org>
5345Date: Sun Feb 5 10:51:57 2006 -0800
5346
5347 Revert "[PATCH] x86_64: Fix the node cpumask of a cpu going down"
5348
5349 This reverts commit 10f4dc8b27ac42f930ac55adb8c521264dc997f8.
5350
5351 Quoth Andi Kleen:
5352 "Kiran decided that it makes the problem worse than it was before.
5353 Fixing it fully requires more work which is too much for 2.6.16. So
5354 please revert that commit for now."
5355
5356 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5357commit 392c14beaca2ee85a98d0c6b453501be67423a20
5358Author: Linus Torvalds <torvalds@g5.osdl.org>
5359Date: Thu Dec 29 13:01:54 2005 -0800
5360
5361 Revert radeon AGP aperture offset changes
5362
5363 This reverts the series of commits
5364
5365 67dbb4ea33731415fe09c62149a34f472719ac1d
5366 281ab031a8c9e5b593142eb4ec59a87faae8676a
5367 47807ce381acc34a7ffee2b42e35e96c0f322e52
5368
5369 that changed the GART VM start offset. It fixed some machines, but
5370 seems to continually interact badly with some X versions.
5371
5372 Quoth Ben Herrenschmidt:
5373
5374 "So I think at this point, the best is that we keep the old bogus code
5375 that at least is consistent with the bug in the server. I'm working on a
5376 big patch to X that reworks the memory map stuff completely and fixes
5377 those issues on the server side, I'll do a DRM patch matching this X fix
5378 as well so that the memory map is only ever set in one place and with
5379 what I hope is a correct algorithm..."
5380
5381 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5382commit 49d7bc64283970ee83d2c954d04ba00d04e5943d
5383Author: Linus Torvalds <torvalds@g5.osdl.org>
5384Date: Mon Dec 12 11:25:04 2005 -0800
5385
5386 Revert revert of "[SCSI] fix usb storage oops"
5387
5388 This reverts commit 1b0997f561bf46689cc6e0903f342e9bf2506bf1, which in
5389 turn reverted 34ea80ec6a02ad02e6b9c75c478c18e5880d6713 (which is thus
5390 re-instated).
5391
5392 Quoth James Bottomley:
5393
5394 "All it's doing is deferring the device_put() from the
5395 scsi_put_command() to after the scsi_run_queue(), which doesn't fix
5396 the sleep while atomic problem of the device release method. In both
5397 cases we still get the semaphore in atomic context problem which is
5398 caused by scsi_reap_target() doing a device_del(), which I assumed
5399 (wrongly) was valid from atomic context."
5400
5401 who also promised to fix scsi_reap_target().
5402
5403 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5404commit 1b0997f561bf46689cc6e0903f342e9bf2506bf1
5405Author: Linus Torvalds <torvalds@g5.osdl.org>
5406Date: Fri Dec 2 15:54:37 2005 -0800
5407
5408 Revert "[SCSI] fix usb storage oops"
5409
5410 This reverts commit 34ea80ec6a02ad02e6b9c75c478c18e5880d6713.
5411
5412 It does a put_device() from softirq context, which is bad since it gets
5413 a semaphore for reading.
5414
5415 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5416commit 346f7dbb17cb7d84317b4410df2e8f7bf2e57f44
5417Author: Linus Torvalds <torvalds@g5.osdl.org>
5418Date: Wed Nov 30 10:22:30 2005 -0800
5419
5420 Revert "[PATCH] pci_ids.h: remove duplicate entries"
5421
5422 This reverts commit c9d6073fb3cda856132dd544d537679f9715436c.
5423
5424 It was totally bogus.
5425
5426 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5427commit f747307ed1defcdfd37a3ef84c48e3138691cd26
5428Author: Linus Torvalds <torvalds@g5.osdl.org>
5429Date: Tue Nov 29 14:21:57 2005 -0800
5430
5431 Revert "[PATCH] drivers/message/fusion/mptbase.c: make code static"
5432
5433 This reverts commit 252ac865535e1ea9cc2d28be83f477d8d8b961a2.
5434
5435 It impacts the LSI customers using the mptstm target mode drivers
5436 (source tar-ball at
5437
5438 ftp://ftp.lsil.com/HostAdapterDrivers/linux/Fusion-MPT/mptstm-1.00.13-src.tar.gz
5439
5440 for those who care).
5441
5442 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5443commit 2d0ebb36038c0626cde662a3b06da9787cfb68c3
5444Author: Linus Torvalds <torvalds@g5.osdl.org>
5445Date: Wed Nov 23 08:44:05 2005 -0800
5446
5447 Revert "[NET]: Shut up warnings in net/core/flow.c"
5448
5449 This reverts commit af2b4079ab154bd12e8c12b02db5f31b31babe63
5450
5451 Changing the #define to an inline function breaks on non-SMP builds,
5452 since wuite a few places in the kernel do not implement the ipi handler
5453 when compiling for UP.
5454
5455 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5456commit 0fde7f591860e48dd210144ff24c061da03bfa39
5457Author: Linus Torvalds <torvalds@g5.osdl.org>
5458Date: Sat Nov 12 10:14:02 2005 -0800
5459
5460 Revert "[PATCH] fbcon: Add rl (Roman Large) font"
5461
5462 This reverts 998e6d51162707685336ff99c029c8911b270d32 commit.
5463commit 1e4c85f97fe26fbd70da12148b3992c0e00361fd
5464Author: Linus Torvalds <torvalds@g5.osdl.org>
5465Date: Mon Oct 31 19:16:17 2005 -0800
5466
5467 Revert "i386: move apic init in init_IRQs"
5468
5469 Commit f2b36db692b7ff6972320ad9839ae656a3b0ee3e causes a bootup hang on
5470 at least one machine. Revert for now until we understand why. The old
5471 code may be ugly, but it works.
5472
5473 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5474commit 79b95a454bb5c1d9b7287d1016a70885ba3f346c
5475Author: Linus Torvalds <torvalds@g5.osdl.org>
5476Date: Thu Oct 27 16:28:39 2005 -0700
5477
5478 Revert "x86-64: Avoid unnecessary double bouncing for swiotlb"
5479
5480 Commit id 6142891a0c0209c91aa4a98f725de0d6e2ed4918
5481
5482 Andi Kleen reports that it seems to break things for some people,
5483 and since it's purely a small optimization, revert it for now.
5484
5485 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5486commit a362f463a6d316d14daed0f817e151835ce97ff7
5487Author: Linus Torvalds <torvalds@g5.osdl.org>
5488Date: Thu Oct 27 09:07:33 2005 -0700
5489
5490 Revert "remove false BUG_ON() from run_posix_cpu_timers()"
5491
5492 This reverts commit 3de463c7d9d58f8cf3395268230cb20a4c15bffa.
5493
5494 Roland has another patch that allows us to leave the BUG_ON() in place
5495 by just making sure that the condition it tests for really is always
5496 true.
5497
5498 That goes in next.
5499
5500 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5501commit 9465bee863bc4c6cf1566c12d6f92a8133e3da5c
5502Author: Linus Torvalds <torvalds@g5.osdl.org>
5503Date: Fri Oct 21 15:36:00 2005 -0700
5504
5505 Revert "Fix cpu timers exit deadlock and races"
5506
5507 Revert commit e03d13e985d48ac4885382c9e3b1510c78bd047f, to be replaced
5508 by a much nicer fix from Roland.
5509commit 6e3254c4e2927c117044a02acf5f5b56e1373053
5510Author: Linus Torvalds <torvalds@g5.osdl.org>
5511Date: Fri Sep 30 12:38:27 2005 -0700
5512
5513 Revert "x86-64: Reverse order of bootmem lists"
5514
5515 As requested by Thomas Gleixner <tglx@linutronix.de>:
5516
5517 "5d3d0f7704ed0bc7eaca0501eeae3e5da1ea6c87 breaks a couple of ARM
5518 boards, which depend on the historical bootmem allocation order.
5519 There is a cleaner solution around to remove the pgdat list
5520 completely, but this is a topic for post 2.6.14
5521
5522 Andi signalled ACK already."
5523
5524 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5525commit 4a8342d233a39ee582e9f7260e12d2f5fd194a05
5526Author: Linus Torvalds <torvalds@g5.osdl.org>
5527Date: Thu Sep 29 15:18:21 2005 -0700
5528
5529 Revert task flag re-ordering, add comments
5530
5531 Roland points out that the flags end up having non-obvious dependencies
5532 elsewhere, so revert aa55a08687059aa169d10a313c41f238c2070488 and add
5533 some comments about why things are as they are.
5534
5535 We'll just have to fix up the broken comparisons. Roland has a patch.
5536
5537 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5538commit c231c7db30faf93419fc22d680f74d816bea70e2
5539Author: Linus Torvalds <torvalds@g5.osdl.org>
5540Date: Wed Aug 17 13:07:28 2005 -0700
5541
5542 Revert unnecessary zlib_inflate/inftrees.c fix
5543
5544 It turns out that empty distance code tables are not an error, and that
5545 a compressed block with only literals can validly have an empty table
5546 and should not be flagged as a data error.
5547
5548 Some old versions of gzip had problems with this case, but it does not
5549 affect the zlib code in the kernel.
5550
5551 Analysis and explanations thanks to Sergey Vlasov <vsu@altlinux.ru>
5552
5553 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5554commit 2ba84684e8cf6f980e4e95a2300f53a505eb794e
5555Author: Linus Torvalds <torvalds@g5.osdl.org>
5556Date: Sun Aug 14 18:21:30 2005 -0700
5557
5558 Revert PCIBIOS_MIN_IO changes for 2.6.13
5559
5560 This reverts commits
5561
5562 71db63acff69618b3d9d3114bd061938150e146b
5563 [PATCH] increase PCIBIOS_MIN_IO on x86
5564
5565 and
5566
5567 0b2bfb4e7ff61f286676867c3508569bea6fbf7a
5568 ACPI: increase PCIBIOS_MIN_IO on x86
5569
5570 since Lukas Sandströ<lukass@etek.chalmers.se> reports that this breaks
5571 his on-board nvidia audio.
5572
5573 We should re-visit this later. For now we revert the change
5574
5575 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5576commit b4b08e581fac8e0ba9ae348bdc13246c9798c99e
5577Author: Linus Torvalds <torvalds@g5.osdl.org>
5578Date: Sun Aug 14 15:43:39 2005 -0700
5579
5580 Revert "dc395x: Fix support for highmem"
5581
5582 It introduces a repeatable oops in the driver, which is a bigger problem
5583 than the patch tries to solve. From the original description:
5584
5585 Author: Jamie Lenehan <lenehan@twibble.org>
5586 Date: Thu Mar 3 14:41:40 2005 +0200
5587
5588 [PATCH] dc395x: Fix support for highmem
5589
5590 From: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
5591
5592 Removes the page_to_virt and maps sg lists dynamically.
5593 This makes the driver work with highmem pages.
5594
5595 Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
5596 Signed-off-by: Jamie Lenehan <lenehan@twibble.org>
5597 Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
5598
5599 Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
5600 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5601commit dc836b5b6fcde95f750a4790d8200fabaf563dc9
5602Author: Linus Torvalds <torvalds@g5.osdl.org>
5603Date: Mon Aug 8 18:46:09 2005 -0700
5604
5605 Revert "[PATCH] PCI: restore BAR values..."
5606
5607 Revert commit fec59a711eef002d4ef9eb8de09dd0a26986eb77, which is
5608 breaking sparc64 that doesn't have a working pci_update_resource.
5609
5610 We'll re-do this after 2.6.13 when we'll do it all properly.
5611commit 697a2d63a3844caaa2b6565ab7f3d69086af94d4
5612Author: Linus Torvalds <torvalds@evo.osdl.org>
5613Date: Mon Aug 1 12:37:54 2005 -0700
5614
5615 Revert ACPI interrupt resume changes
5616
5617 If there are devices that use interrupts over a suspend event, ACPI must
5618 restore the PCI interrupt links on resume. Anything else breaks any
5619 device that hasn't been converted to the new (dubious) PM rules.
5620
5621 Drivers that need the irq free/re-aquire sequence can be done one by one
5622 independently of this one.
5623commit 889371f61fd5bb914d0331268f12432590cf7e85
5624Author: Linus Torvalds <torvalds@g5.osdl.org>
5625Date: Sat Jul 30 13:41:56 2005 -0700
5626
5627 Revert "yenta free_irq on suspend"
5628
5629 ACPI is wrong. Devices should not release their IRQ's on suspend and
5630 re-aquire them on resume. ACPI should just re-init the IRQ controller
5631 instead of breaking most drivers very subtly.
5632
5633 Breakage reported by Hugh Dickins <hugh@veritas.com>
5634
5635 Undo: d8c4b4195c7d664baf296818bf756775149232d3
5636
5637 Signed-off-by: Linus Torvalds <torvalds@osdl.org>
5638commit 2ac6608c41f8c45371ea9dddae7f99bc2c15d5cf
5639Author: Linus Torvalds <torvalds@g5.osdl.org>
5640Date: Thu Jul 28 10:34:47 2005 -0700
5641
5642 Revert broken "statement with no effect" warning fix
5643
5644 It may shut up gcc, but it also incorrectly changes the semantics of the
5645 smp_call_function() helpers.
5646
5647 You can fix the warning other ways if you are interested (create another
5648 inline function that takes no arguments and returns zero), but
5649 preferably gcc just shouldn't complain about unused return values from
5650 statement expressions in the first place.