So the size of the -rc patches is finally starting to shrink, but we still
have way too many outstanding regression reports. Some of them should be
already fixed in -rc5, but many won't be. So reporters, please do keep
updating your entries as to whether they are fixed or not.
This time around, about 60% of the diffs are in arch code, but that's a
bit misleading, because over half of that is actually just the defconfig
files for SH. With the patch-queues shrinking, big trivial patches like
that easily end up being inordinately noticeable just because there's a
lot of lines there.
But apart from SH, the ia64 and SCSI updates do stand out. The rest is
really pretty randomly spread out. I suspect the shortlog does as good a
job as can be done (and is now small enough to be posted), but here's the
dirstat anyway since I've given it for previous -rc releases.
15.1% arch/ia64/kernel/
18.3% arch/ia64/
2.3% arch/sh/boards/renesas/sh7710voipgw/
2.3% arch/sh/boards/renesas/
2.9% arch/sh/boards/
32.8% arch/sh/configs/
37.6% arch/sh/
59.3% arch/
7.7% drivers/scsi/aic94xx/
2.0% drivers/scsi/libsas/
17.1% drivers/scsi/
21.3% drivers/
2.4% fs/
2.9% include/linux/
3.9% include/
2.7% kernel/
4.4% security/selinux/
5.7% security/
I'd like developers and testers alike to take a good look at the
regression email that Rafael sends out, and please help him keep the
entries up-to-date.
Linus
---
Adrian Bunk (4):
UBI: mtd/ubi/vtbl.c: fix memory leak
[CRYPTO] digest: Include internal.h for prototypes
sh/mm/pg-sh7705.c must #include <linux/fs.h>
sh_ksyms_32.c update for gcc 4.3
Alan Cox (1):
pata_hpt*, pata_serverworks: fix UDMA masking
Andrew Morton (1):
sh: export copy-page() to modules
Andrew Vasquez (4):
[SCSI] qla2xxx: Correct needless clean-up resets during shutdown.
[SCSI] qla2xxx: Correct discrepancies during OVERRUN handling on FWI2-capable cards.
[SCSI] qla2xxx: Correct usage of inconsistent timeout values while issuing ELS commands.
[SCSI] qla2xxx: Update version number to 8.02.00-k9.
Artem Bityutskiy (1):
UBI: fix error message
Aurelien Jarno (1):
x86: clear DF before calling signal handler
Bartlomiej Zolnierkiewicz (2):
ide-cd: mark REQ_TYPE_ATA_PC write requests with REQ_RW flag
ide: fix enabling DMA on it821x in "smart" mode
Bjorn Helgaas (2):
[IA64] remove CONFIG_SMP ifdef in ia64_send_ipi()
[IA64] use dev_printk in video quirk
Boaz Harrosh (2):
[SCSI] gdth: bugfix for the at-exit problems
[SCSI] gdth: fix to internal commands execution
Bryan Wu (1):
[Blackfin] arch: to kill syscalls missing warning by adding new timerfd syscalls
Christian Borntraeger (1):
[S390] sclp_vt220: speed up console output for interactive work
Christoph Lameter (2):
slub statistics: Fix check for DEACTIVATE_REMOTE_FREES
slab numa fallback logic: Do not pass unfiltered flags to page allocator
Dave Jones (3):
[CPUFREQ] Fix missing cpufreq_cpu_put() call in ->show
[CPUFREQ] Fix missing cpufreq_cpu_put() call in ->store
[CPUFREQ] Remove debugging message from e_powersaver
David Chinner (2):
[XFS] 977545 977545 977545 977545 977545 977545 xfsaild causing too many
[XFS] fix inode leak in xfs_iget_core()
David Howells (1):
ntp: use unsigned input for do_div()
David Rientjes (1):
cpusets: fix obsolete comment
David Woodhouse (1):
Really unexport asm/page.h
Dhaval Giani (1):
sched: don't allow rt_runtime_us to be zero for groups having rt tasks
Eric Paris (2):
LSM/SELinux: Interfaces to allow FS to control mount options
NFS: use new LSM interfaces to explicitly set mount options
FUJITA Tomonori (7):
[SCSI] ps3rom: fix wrong resid calculation bug
[SCSI] ps3rom: disable clustering
[SCSI] ibmvstgt: set up scsi_host properly before __scsi_alloc_queue
[SCSI] tgt: stop zero'ing scsi_cmnd
[SCSI] tgt: set the data length properly
[SCSI] tgt: fix build errors when dprintk is defined
parisc: fix IOMMU's device boundary overflow bug on 32bits arch
Gregory Haskins (1):
cpu hotplug: adjust root-domain->online span in response to hotplug event
Harvey Harrison (4):
UBI: fix sparse errors in ubi.h
sh: replace remaining __FUNCTION__ occurrences
[IA64] remove remaining __FUNCTION__ occurrences
[IA64] kprobes arch consolidation build fix
Heiko Carstens (3):
[S390] idle: Fix machine check handling in idle loop.
[S390] Get rid of memcpy gcc warning workaround.
[S390] Fix IPL from NSS.
Herbert Xu (3):
[CRYPTO] skcipher: Move chainiv/seqiv into crypto_blkcipher module
[CRYPTO] authenc: Add missing Kconfig dependency on BLKCIPHER
[CRYPTO] skcipher: Fix section mismatches
Hongjie Yang (1):
[S390] incorrect reipl nss name.
Ingo Molnar (2):
x86: re-add reboot fixups
drivers/char/esp.c: fix bootup lockup
Itaru Kitayama (1):
slub: fix typo in Documentation/vm/slub.txt
Ivan Kokshaysky (1):
alpha: fix iommu-related boot panic
James Bottomley (6):
[SCSI] mvsas: remove execute permission from file
[SCSI] libsas: export sas_find_local_phy function
[SCSI] aic94xx: plumb in I_T_nexus_reset task management function
[SCSI] libsas: use the supplied address for SATA devices rather than changing it
[SCSI] libsas: misc fixes to the eh path
[SCSI] aic94xx: fix TMF ascb handling to prevent sequencer panic
Jan Beulich (2):
x86: fix merge mistake in i387.c
x86: fix typo in step.c
Jeff Garzik (1):
ahci: work around ATI SB600 h/w quirk
Jesper Juhl (1):
Do not include linux/backing-dev.h twice
Joe Korty (1):
slab: NUMA slab allocator migration bugfix
Joe Perches (1):
slab - use angle brackets for include of kmalloc_sizes.h
Joy Latten (1):
[CRYPTO] xcbc: Fix crash with IPsec
KOSAKI Motohiro (1):
[IA64] remove unnecessary nfs includes from sys_ia32.c
Karsten Keil (1):
bluetooth: Add another Broadcom device
Karsten Wiese (1):
time: don't touch an offlined CPU's ts->tick_stopped in tick_cancel_sched_timer()
Ke Wei (1):
[SCSI] mvsas: fix phy sas address
Kristoffer Ericson (1):
sh: hp6xx: Correct APM output.
Krzysztof Oledzki (1):
[SCSI] mpt fusion: don't oops if NumPhys==0
Linus Torvalds (1):
Linux 2.6.25-rc5
Martin Schwidefsky (1):
[S390] Update default configuration.
Miao Xie (1):
sched: fix the wrong time slice value for SCHED_FIFO tasks
Michael Hennerich (1):
[Blackfin] arch: fix bug - allow SDH driver to be used as module
Michael Holzheu (1):
[S390] Load disabled wait psw if reipl fails.
Mike Christie (4):
[SCSI] iscsi regression: check for zero max session cmds
[SCSI] qla4xxx: fix host reset dpc race
[SCSI] qla4xxx: regression - add start scan callout
[SCSI] iscsi class: regression - fix races with state manipulation and blocking/unblocking
Mike Frysinger (2):
[Blackfin] arch: fix atomic and32/xor32 comments and ENDPROC markings
[Blackfin] arch: current_l1_stack_save is a pointer, so use NULL rather than 0
Nick Cheng (1):
[SCSI] arcmsr: update version and changelog
Nick Piggin (1):
slub: Do not cross cacheline boundaries for very small objects
Paul Mundt (8):
sh: Fix up section mismatches.
fb: pvr2fb: Fix up remaining section mismatch.
sh: hp6xx: Fix up hp6xx_apm build failure.
fb: hitfb: Balance probe/remove section annotations.
sh: Update r7780mp defconfig.
sh: Flag PMB support as EXPERIMENTAL.
sh: Fix up SH7710 VoIP-GW build.
sh: Fix up the sh64 build.
Pavel Roskin (1):
sched: export task_nice
Peter Korsgaard (1):
x86-boot: don't request VBE2 information
Peter Teoh (1):
ide: fix buggy code in ide_register_hw()
Peter Zijlstra (2):
sched: retain vruntime
sched: rt-group: fixup schedulability constraints calculation
Petr Tesarik (6):
[IA64] access user RBS directly
[IA64] do not sync RBS when changing PT_AR_BSP or PT_CFM
[IA64] remove find_thread_for_addr()
[IA64] convert sys_ptrace to arch_ptrace
[IA64] remove duplicate code from arch_ptrace()
[IA64] arch_ptrace() cleanup
Ralph Wuerthner (1):
[S390] zcrypt: fix ap_device_list handling
Randy Dunlap (4):
[SCSI] docbook: fix scsi source file
[SCSI] docbook: fix fusion source files
ide: move ide.txt to Documentation/ide/
ide: update references to Documentation/ide/ide.txt (v2)
Roland McGrath (2):
x86_64: make ptrace always sign-extend orig_ax to 64 bits
Fix waitid si_code regression
Roman Zippel (1):
time: remove obsolete CLOCK_TICK_ADJUST
S.Çağlar Onur (1):
UBI: silence warning
Sam Ravnborg (1):
[CPUFREQ] fix section mismatch warnings
Sebastian Siewior (1):
[CRYPTO] xts: Use proper alignment
Segher Boessenkool (1):
time: prevent the loop in timespec_add_ns() from being optimised away
Simon Horman (1):
[IA64] update efi region debugging to use MB, GB and TB as well as KB
Stefan Weinhuber (2):
[S390] dasd: let dasd erp matching recognize alias recovery
[S390] dasd: fix reference counting in display method for proc/dasd/devices
Steven Rostedt (1):
sched: balance RT task resched only on runqueue
Tilman Schmidt (1):
gigaset: fix Oops on module unload regression
Tom Talpey (1):
SUNRPC: Fix a nfs4 over rdma transport oops
Trond Myklebust (3):
NFS: Fix an f_mode/f_flags confusion in fs/nfs/write.c
NFS: Fix the fsid revalidation in nfs_update_inode()
NFS: Fix dentry revalidation for NFSv4 referrals and mountpoint crossings
goda.yusuke (1):
sh: update se7780 defconfig
i see the -rc5 patches are uploaded, but the git tree isn't updated.
charles
On Sun, 9 Mar 2008, charles gagalac wrote:
>
> i see the -rc5 patches are uploaded, but the git tree isn't updated.
Duh. Thanks for pointing that out. It's uptodate now (it wasn't really far
off before either - just the tag and one really trivial patch to remove a
duplicate header include)
Linus
On Mon, 10 Mar 2008, Alan Cox wrote:
>
> > NAME = Funky Weasel is Jiggy wit it
>
> Typo ??
The "wit"? No. Google for "jiggy wit it".
Linus
On 10-03-08 15:58, Alan Cox wrote:
>> NAME = Funky Weasel is Jiggy wit it
>
> Typo ??
Oh you... old person.
http://en.wikipedia.org/wiki/Gettin'_Jiggy_Wit_It
Rene.
> I'd like developers and testers alike to take a good look at the
> regression email that Rafael sends out, and please help him keep the
> entries up-to-date.
figured out another ACPI related regression today.
randconfig testing triggered an early boot-time hang on a laptop of mine
(32-bit x86, config attached) - the screen was scrolling ACPI AML
exceptions [with no serial port and no early debugging available].
v2.6.24 works fine on that laptop with the same .config, so after a few
hours of bisection (had to restart it 3 times - other regressions
interacted), it honed in on this commit:
| 10270d4838bdc493781f5a1cf2e90e9c34c9142f is first bad commit
|
| Author: Linus Torvalds <[email protected]>
| Date: Wed Feb 13 09:56:14 2008 -0800
|
| acpi: fix acpi_os_read_pci_configuration() misuse of raw_pci_read()
reverting this commit ontop of -rc5 gave a correctly booting kernel.
but this commit fixes a real bug so the real question is, why did it
break the bootup?
after quite some head-scratching, the following change stood out:
- pci_id->bus = tu8;
+ pci_id->bus = val;
pci_id->bus is defined as u16:
struct acpi_pci_id {
u16 segment;
u16 bus;
...
and 'tu8' changed from u8 to u32. So previously we'd unconditionally
mask the return value of acpi_os_read_pci_configuration()
(raw_pci_read()) to 8 bits, but now we just trust whatever comes back
from the PCI access routines and only crop it to 16 bits.
But if the high 8 bits of that result contains any noise then we'll
write that into ACPI's PCI ID descriptor and confuse the heck out of the
rest of ACPI.
So lets check the PCI-BIOS code on that theory. We have this codepath
for 8-bit accesses (arch/x86/pci/pcbios.c:pci_bios_read()):
switch (len) {
case 1:
__asm__("lcall *(%%esi); cld\n\t"
"jc 1f\n\t"
"xor %%ah, %%ah\n"
"1:"
: "=c" (*value),
"=a" (result)
: "1" (PCIBIOS_READ_CONFIG_BYTE),
"b" (bx),
"D" ((long)reg),
"S" (&pci_indirect));
aha! The "=a" output constraint puts the full 32 bits of EAX into
*value. But if the BIOS's routines set any of the high bits to nonzero,
we'll return a value with more set in it than intended.
the other, more common PCI access methods (v1 and v2 PCI reads) clear
out the high bits already, for example pci_conf1_read() does:
switch (len) {
case 1:
*value = inb(0xCFC + (reg & 3));
which explicitly converts the return byte up to 32 bits and zero-extends
it.
so zero-extending the result in the PCI-BIOS read routine fixes the
regression on my laptop. ( It might fix some other long-standing issues
we had with PCI-BIOS during the past decade ... ) Both 8-bit and 16-bit
accesses were buggy.
Signed-off-by: Ingo Molnar <[email protected]>
---
arch/x86/pci/pcbios.c | 10 ++++++++++
1 file changed, 10 insertions(+)
Index: linux/arch/x86/pci/pcbios.c
===================================================================
--- linux.orig/arch/x86/pci/pcbios.c
+++ linux/arch/x86/pci/pcbios.c
@@ -198,6 +198,11 @@ static int pci_bios_read(unsigned int se
"b" (bx),
"D" ((long)reg),
"S" (&pci_indirect));
+ /*
+ * Zero-extend the result beyond 8 bits, do not trust the
+ * BIOS having done it:
+ */
+ *value &= 0xff;
break;
case 2:
__asm__("lcall *(%%esi); cld\n\t"
@@ -210,6 +215,11 @@ static int pci_bios_read(unsigned int se
"b" (bx),
"D" ((long)reg),
"S" (&pci_indirect));
+ /*
+ * Zero-extend the result beyond 16 bits, do not trust the
+ * BIOS having done it:
+ */
+ *value &= 0xffff;
break;
case 4:
__asm__("lcall *(%%esi); cld\n\t"
On Mon, 10 Mar 2008, Ingo Molnar wrote:
>
> so zero-extending the result in the PCI-BIOS read routine fixes the
> regression on my laptop. ( It might fix some other long-standing issues
> we had with PCI-BIOS during the past decade ... ) Both 8-bit and 16-bit
> accesses were buggy.
Oh, wow. That must have been there forever, but very few people probably
ever cared.
And why is your laptop using the BIOS routines anyway? Or was that just a
result of your randconfig having turned off the sane config access
routines?
Linus
* Linus Torvalds <[email protected]> wrote:
> On Mon, 10 Mar 2008, Ingo Molnar wrote:
> >
> > so zero-extending the result in the PCI-BIOS read routine fixes the
> > regression on my laptop. ( It might fix some other long-standing
> > issues we had with PCI-BIOS during the past decade ... ) Both 8-bit
> > and 16-bit accesses were buggy.
>
> Oh, wow. That must have been there forever, but very few people
> probably ever cared.
>
> And why is your laptop using the BIOS routines anyway? Or was that
> just a result of your randconfig having turned off the sane config
> access routines?
yeah, randconfig generated this:
CONFIG_PCI=y
CONFIG_PCI_GOBIOS=y
# CONFIG_PCI_GOMMCONFIG is not set
# CONFIG_PCI_GODIRECT is not set
# CONFIG_PCI_GOANY is not set
CONFIG_PCI_BIOS=y
which forced the PCI code into the BIOS access method on this otherwise
modern system. I disable some of the really-known-to-be-broken .config
options in randconfig - but i still bravely keept CONFIG_GO_PCIBIOS in
the randomized space, which triggered this.
On another box, which has a different BIOS, this bug never happened, in
tens of thousands of bootup tests.
Ingo
On Monday 10 March 2008, Ingo Molnar wrote:
>
> * Linus Torvalds <[email protected]> wrote:
>
> > On Mon, 10 Mar 2008, Ingo Molnar wrote:
> > >
> > > so zero-extending the result in the PCI-BIOS read routine fixes the
> > > regression on my laptop. ( It might fix some other long-standing
> > > issues we had with PCI-BIOS during the past decade ... ) Both 8-bit
> > > and 16-bit accesses were buggy.
> >
> > Oh, wow. That must have been there forever, but very few people
> > probably ever cared.
> >
> > And why is your laptop using the BIOS routines anyway? Or was that
> > just a result of your randconfig having turned off the sane config
> > access routines?
>
> yeah, randconfig generated this:
>
> CONFIG_PCI=y
> CONFIG_PCI_GOBIOS=y
> # CONFIG_PCI_GOMMCONFIG is not set
> # CONFIG_PCI_GODIRECT is not set
> # CONFIG_PCI_GOANY is not set
> CONFIG_PCI_BIOS=y
>
> which forced the PCI code into the BIOS access method on this otherwise
> modern system. I disable some of the really-known-to-be-broken .config
> options in randconfig - but i still bravely keept CONFIG_GO_PCIBIOS in
> the randomized space, which triggered this.
>
> On another box, which has a different BIOS, this bug never happened, in
> tens of thousands of bootup tests.
I think this re-enforces three Axioms
1. Ingo rules!
2. Even the most obvious fix carries some risk
3. The more config options we maintain, the more fragmented is our test coverage.
thanks Ingo,
-Len