]> git.hungrycats.org Git - linux/log
linux
22 years agoLinux v2.5.54 v2.5.54
Linus Torvalds [Wed, 1 Jan 2003 11:11:40 +0000 (03:11 -0800)]
Linux v2.5.54

22 years agoRemove old kernel version test that didn't even compile.
Linus Torvalds [Wed, 1 Jan 2003 11:06:10 +0000 (03:06 -0800)]
Remove old kernel version test that didn't even compile.

Noted by Steven Barnhart <sbarn03@softhome.net>

22 years ago[PATCH] Modules 3/3: Sort sections
Rusty Russell [Wed, 1 Jan 2003 11:02:38 +0000 (03:02 -0800)]
[PATCH] Modules 3/3: Sort sections

RTH's final complaint (so far 8) was that we should sort the module
sections: archs might require some sections to be adjacent, so they can
all be reached by a relative pointer (ie.  GOT pointer).  This
implements that reordering, and simplfies the module interface for
architectures as well.

Previously an arch could specify it wanted extra space, but not where
that space would be.  The new method (used only by PPC so far) is to
allocate an empty section (in asm/module.h or by setting LDFLAGS_MODULE
to use an arch specific linker script), and expand that to the desired
size in "module_frob_arch_sections()".

22 years ago[PATCH] Modules 2/3: Use sh_addr instead of sh_offset
Rusty Russell [Wed, 1 Jan 2003 11:02:32 +0000 (03:02 -0800)]
[PATCH] Modules 2/3: Use sh_addr instead of sh_offset

From: Richard Henderson <rth@twiddle.net>

Richard points out that we should be using sh_addr to hold the address,

The original patch used to overload sh_offset to a pointer to the
location of the section.  This uses sh_addr, which is more correct
and less surprising.

22 years ago[PATCH] Modules 1/3: remove common section handling
Rusty Russell [Wed, 1 Jan 2003 11:02:25 +0000 (03:02 -0800)]
[PATCH] Modules 1/3: remove common section handling

As RTH pointed out, we use -fno-common for the kernel (otherwise we'd
have to sort out the small symbols anyway).

So the common section allocation in the module code is pointless.

22 years agoIgnore generated files in lib/
Linus Torvalds [Wed, 1 Jan 2003 09:19:06 +0000 (01:19 -0800)]
Ignore generated files in lib/

22 years agoFrom louis.zhuang@intel.com: missed field_width reset in vsscanf()
Linus Torvalds [Wed, 1 Jan 2003 09:11:11 +0000 (01:11 -0800)]
From louis.zhuang@intel.com: missed field_width reset in vsscanf()

22 years ago[PATCH] quota locking update
Andrew Morton [Wed, 1 Jan 2003 08:46:22 +0000 (00:46 -0800)]
[PATCH] quota locking update

I've been carrying this since Jan sent it out a month or two ago.
I don't know if anyone has tested it though.  The sort of people
who use quotas tend to like nice stable kernels.  I read through it,
but can't say that I know enough about quotas to know if it makes sense.
The wait_on_dquot() synchronisation is a bit odd.

I do need to do a round of stability testing with this and ext3 - the
interaction between quotas and ext3 is an area where we've had deadlocks
in the past.

But the quota locking is definitely looking crufty, and I'd suggest that
we run with this..

Patch from Jan Kara <jack@suse.cz>

"I'm resending you the patch with new quota SMP locking.  The patch removes
 BKL and replaces it with two spinlocks protecting quota lists and data
 stored in dquot structures.  Also non-SMP locking was changed a bit make SMP
 locking easier (eg.  we got rid of not very nice dq_dup_ref counters).  The
 patch is against 2.5.48 but applies well also to 2.5.49.  Would you please
 apply the patch?"

- Change dqoff_sem from a semaphore to an rwsem.

- Convert dqi_flags from an int to a ulong and use test_bit/set_bit rather
  thatn &/|

- The various exported quota operations now run without lock_kernel().
  This means that things like DQUOT_ALLOC_SPACE no longer take lock_kernel()
  in out high-perfomance filesystems.  Nice.

- Replace lock_kernel() in the quota code with two quota-private global
  locks.

- Replace all the open-coded waitqueue management with a semaphore
  (wait_on_dquot())

22 years ago[PATCH] some ppp/usecount bug
Christoph Hellwig [Wed, 1 Jan 2003 07:43:32 +0000 (23:43 -0800)]
[PATCH] some ppp/usecount bug

There's a bunch of useless MOD_INC_USE_COUNT/MOD_DEC_USE_COUNT in
ppp-related code when it manipulates the use count in functions
that are only called from other modules, i.e. it's locked in by
the depencies anyway.  Get rid of it.

22 years ago[PATCH] devfs creptomancy
Christoph Hellwig [Wed, 1 Jan 2003 07:32:55 +0000 (23:32 -0800)]
[PATCH] devfs creptomancy

As already state in the thread about Adam's devfs reimplementation there
is much devfs functionality that is unused or only used by the
arch/ia64/sn/ code that currently doesn't even compile in 2.5 and that
will get it's own filesystem that fits the needs better when SGI moves
to 2.6.

(the first hunk is the only exception to the above rule, but it's just
a debug printk :))

22 years ago[PATCH] include <linux/vfs.h> only in files actually needing it
Christoph Hellwig [Wed, 1 Jan 2003 04:07:20 +0000 (20:07 -0800)]
[PATCH] include <linux/vfs.h> only in files actually needing it

fs.h only needs the forward-declaration of struct statfs

22 years ago[PATCH] i810: get rid of a forgotten Rules.make include
Tomas Szepe [Wed, 1 Jan 2003 04:06:34 +0000 (20:06 -0800)]
[PATCH] i810: get rid of a forgotten Rules.make include

22 years ago[PATCH] Fusion-MPT Update (2.03.01.01)
Pam Delaney [Wed, 1 Jan 2003 03:56:26 +0000 (19:56 -0800)]
[PATCH] Fusion-MPT Update (2.03.01.01)

This upgrades the Fusion-MPT driver from 2.03.00.02 to 2.03.01.01.

Bug Fixes:
 o Added back missing queuecommand entry point define ?!
 o Added to code to break marriage of two controllers during unload
  (could cause a panic)
 o SCSI driver will de-register with base driver if no SCSI-capable
   adapters found

Minor Changes:
 o Removed errant spaces at ends of lines  (most of the changes)
 o Moved code around (and in-lined) some functions for performance reasons.
 o Modified /proc functionality to facilitate testing with 2.5
 o Added a call to synchronize_irq on unload (HP request)
 o Modified load of base to close a potential hole
 o Added code to set the FW IO coalescing depth (IBM request)
 o Changed return when mptctl driver registration fails (Kernel.org request)
 o SCSI driver detect routine calls a generic spinlock for all kernels
   (Kernel.org request)
 o Controller RAID page dynamic instead of static

Currently running a multi-disk stress test w/ 2.5.53,  this patch and driver
built-in. Verified basic reset handling is working properly.

22 years agoShowing the logo on every open became annoying. I leave it up to the driver writers...
James Simmons [Tue, 31 Dec 2002 10:09:03 +0000 (02:09 -0800)]
Showing the logo on every open became annoying. I leave it up to the driver writers when they want to display the logo. Fo embeded devices it probably is wise to have them set there mode themselves then display the logo to let them know if the hartdware worked. FOnt header file location changed.

22 years agoSynced to Linus tree,
James Simmons [Tue, 31 Dec 2002 09:59:18 +0000 (01:59 -0800)]
Synced to Linus tree,

22 years agoQuoth James: "Oops, mea culpa on that one. It's missing a trailing `__'
Linus Torvalds [Tue, 31 Dec 2002 07:10:20 +0000 (23:10 -0800)]
Quoth James: "Oops, mea culpa on that one.  It's missing a trailing `__'
on the end of __GNUC_MINOR"

22 years ago[PATCH] __deprecated requires gcc 3.1
Robert Love [Tue, 31 Dec 2002 06:32:18 +0000 (22:32 -0800)]
[PATCH] __deprecated requires gcc 3.1

James Bottomley confirmed the "deprecated" attribute requires gcc 3.1
and onward, not gcc 3.0.

This updates the check in compiler.h accordingly.

22 years agoRemove old module crap.
James Simmons [Tue, 31 Dec 2002 04:41:26 +0000 (20:41 -0800)]
Remove old module crap.

22 years agoMerge
James Simmons [Tue, 31 Dec 2002 04:25:50 +0000 (20:25 -0800)]
Merge

22 years agoMerge bk://are.twiddle.net/tga-2.5
Linus Torvalds [Tue, 31 Dec 2002 04:09:19 +0000 (20:09 -0800)]
Merge bk://are.twiddle.net/tga-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoFind the QT libs in more places.. (At least RH-8.1)
Linus Torvalds [Tue, 31 Dec 2002 03:22:43 +0000 (19:22 -0800)]
Find the QT libs in more places.. (At least RH-8.1)

22 years agoMerge http://linux-isdn.bkbits.net/linux-2.5.make
Linus Torvalds [Tue, 31 Dec 2002 03:06:08 +0000 (19:06 -0800)]
Merge http://linux-isdn.bkbits.net/linux-2.5.make
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoMerge bk://linux-dj.bkbits.net/agpgart
Linus Torvalds [Tue, 31 Dec 2002 03:05:10 +0000 (19:05 -0800)]
Merge bk://linux-dj.bkbits.net/agpgart
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoMerge tetrachloride.(none):/mnt/stuff/kernel/2.5/bk-linus
Dave Jones [Tue, 31 Dec 2002 09:19:00 +0000 (08:19 -0100)]
Merge tetrachloride.(none):/mnt/stuff/kernel/2.5/bk-linus
into tetrachloride.(none):/mnt/stuff/kernel/2.5/agpgart

22 years agoMerge bk://linux-dj.bkbits.net/warnings
Linus Torvalds [Tue, 31 Dec 2002 03:03:13 +0000 (19:03 -0800)]
Merge bk://linux-dj.bkbits.net/warnings
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoMerge tetrachloride.(none):/mnt/stuff/kernel/2.5/bk-linus
Dave Jones [Tue, 31 Dec 2002 08:25:40 +0000 (07:25 -0100)]
Merge tetrachloride.(none):/mnt/stuff/kernel/2.5/bk-linus
into tetrachloride.(none):/mnt/stuff/kernel/2.5/warnings

22 years ago[PATCH] Atari NCR5380 SCSI: bitops operate on long
Geert Uytterhoeven [Tue, 31 Dec 2002 03:01:43 +0000 (19:01 -0800)]
[PATCH] Atari NCR5380 SCSI: bitops operate on long

Atari NCR5380 SCSI driver tag bitmap updates:
  - Use DECLARE_BITMAP() to declare the tag bitmap
  - Remove `MAX_TAGS must be a multiple of 32', which is no longer true
  - Declare and use CLEAR_BITMAP() to set all bits in a bitmap to zero
  - Fix bitops call problems that got unnoticed before

22 years ago[PATCH] fix up UP-APIC compile
Tomas Szepe [Tue, 31 Dec 2002 02:59:00 +0000 (18:59 -0800)]
[PATCH] fix up UP-APIC compile

22 years agoFix up numerous '`xxxxx' is not at beginning of declaration' style warnings.
Dave Jones [Tue, 31 Dec 2002 02:28:13 +0000 (01:28 -0100)]
Fix up numerous '`xxxxx' is not at beginning of declaration' style warnings.

22 years ago[AGP] mmap readability cleanup
Dave Jones [Tue, 31 Dec 2002 01:06:40 +0000 (00:06 -0100)]
[AGP] mmap readability cleanup

22 years ago[AGP] Gratuitous whitespace cleanups
Dave Jones [Tue, 31 Dec 2002 01:01:25 +0000 (00:01 -0100)]
[AGP] Gratuitous whitespace cleanups

22 years ago[AGP] Clean up atomic usage of agp_bridge.agp_in_use
Dave Jones [Tue, 31 Dec 2002 00:50:58 +0000 (23:50 -0100)]
[AGP] Clean up atomic usage of agp_bridge.agp_in_use

22 years ago[AGP] Remove broken deprecated module locking.
Dave Jones [Tue, 31 Dec 2002 00:49:37 +0000 (23:49 -0100)]
[AGP] Remove broken deprecated module locking.

22 years agoMerge tp1.ruhr-uni-bochum.de:/home/kai/kernel/v2.5/linux-2.5
Kai Germaschewski [Mon, 30 Dec 2002 19:32:59 +0000 (13:32 -0600)]
Merge tp1.ruhr-uni-bochum.de:/home/kai/kernel/v2.5/linux-2.5
into tp1.ruhr-uni-bochum.de:/scratch/kai/linux-2.5.make

22 years agokbuild: Provide "make some/dir/module.ko"
Kai Germaschewski [Mon, 30 Dec 2002 19:21:46 +0000 (13:21 -0600)]
kbuild: Provide "make some/dir/module.ko"

Now that we have a unique suffix for kernel modules (.ko), we
can extend to former "make some/object.o" to work for kernel modules
as well, even composite ones.

It's as easy as "make some/dir/module.ko".

(Sam Ravnborg/me)

22 years agokbuild/x86_64: archhelp, $(build) usage and cleaning
Sam Ravnborg [Mon, 30 Dec 2002 19:16:30 +0000 (13:16 -0600)]
kbuild/x86_64: archhelp, $(build) usage and cleaning

Moved archhelp to arch/x86_64/Makefile
Introduced usage of $(build) and $(clean)
Use kbuild clean infrastructure

22 years agokbuild/sparc64: archhelp and $(build)
Sam Ravnborg [Mon, 30 Dec 2002 19:14:42 +0000 (13:14 -0600)]
kbuild/sparc64: archhelp and $(build)

Moved archhelp to arch/sparc64/Makefile
introduced usage of $(build)
Removed superflous targets archclean and archmrproper

22 years agokbuild/arm: archhelp and $(build)
Sam Ravnborg [Mon, 30 Dec 2002 19:13:36 +0000 (13:13 -0600)]
kbuild/arm: archhelp and $(build)

Moved archhelp to arch/arm/Makefile
Introduced usage of $(build) and $(clean)

22 years agokbuild: Move archhelp to arch/$(ARCH)/Makefile
Sam Ravnborg [Mon, 30 Dec 2002 19:11:46 +0000 (13:11 -0600)]
kbuild: Move archhelp to arch/$(ARCH)/Makefile

arch/$(ARCH)/Makefile already contains the kbuild required additions
to allow the kernel to be built for the architecture in question.
Moving archhelp centralise this information, and no longer require a
boot directory to exist to utilise this feature.

Update i386 to define archhelp in arch/$(ARCH)/Makefile
Other architectures will be updated in next cset.

22 years agokbuild: $(build) and $(clean) macros for make invocation
Sam Ravnborg [Mon, 30 Dec 2002 19:09:26 +0000 (13:09 -0600)]
kbuild: $(build) and $(clean) macros for make invocation

The former macro $(descend ...) hide for make the fact that a recursively make was
invoked. The replacement $(Q)$(MAKE) -f scripts/Makefile.build obj=dir was too verbose.

Introduced $(build) and $(clean) allowing the following syntax:
$(Q)$(MAKE) $(build)=arch/i386/boot target
and similar for clean.

Introduced $(build) and $(clean) in general, and for i386 architecture.

22 years agokbuild: More src/objtree fixes
Sam Ravnborg [Mon, 30 Dec 2002 19:06:48 +0000 (13:06 -0600)]
kbuild: More src/objtree fixes

Add $(srctree)/ etc. to some more places where it was still missing.

22 years agoMerge are.twiddle.net:/home/rth/BK/linus-2.5
Richard Henderson [Mon, 30 Dec 2002 16:32:49 +0000 (08:32 -0800)]
Merge are.twiddle.net:/home/rth/BK/linus-2.5
into are.twiddle.net:/home/rth/BK/tga-2.5

22 years ago[TGAFB] Implement the fb_imageblit hook.
Richard Henderson [Mon, 30 Dec 2002 16:23:48 +0000 (08:23 -0800)]
[TGAFB] Implement the fb_imageblit hook.

Speeds up rendering of text by around 7x for 8bpp cards,
as you'd expect from the difference in the volume of data
passed across the bus.  Thus the win should be about 31x
for 32bpp cards.

22 years agokbuild: Fix kallsyms on 64 bit archs
Kai Germaschewski [Mon, 30 Dec 2002 16:14:32 +0000 (10:14 -0600)]
kbuild: Fix kallsyms on 64 bit archs

The generated .tmp_kallsyms.S needs to use .long / .quad for
32/64 bit archs. To know which arch we're compiling for, we
use the preprocessor and BITS_PER_LONG from <asm/types.h>.

Unfortunately, asm/types.h was not safe to include from assembler
files, so lots of #ifndef __ASSEMBLY__ needed to be added - should
be fine now, untested for != i386, though.

22 years agoMerge bk://linuxusb.bkbits.net/linus-2.5
Linus Torvalds [Mon, 30 Dec 2002 14:54:01 +0000 (06:54 -0800)]
Merge bk://linuxusb.bkbits.net/linus-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoUSB: convert more dbg() calls to dev_dbg for the usb core
Greg Kroah-Hartman [Mon, 30 Dec 2002 14:46:52 +0000 (06:46 -0800)]
USB: convert more dbg() calls to dev_dbg for the usb core

22 years agoUSB: convert more dbg() calls to dev_dbg for the ohci driver
Greg Kroah-Hartman [Mon, 30 Dec 2002 14:44:47 +0000 (06:44 -0800)]
USB: convert more dbg() calls to dev_dbg for the ohci driver

22 years agoMerge
Linus Torvalds [Mon, 30 Dec 2002 10:29:20 +0000 (02:29 -0800)]
Merge

22 years agoBump aic7xxx and aic79xx driver versions to reflect recent DV changes.
Justin T. Gibbs [Mon, 30 Dec 2002 12:01:20 +0000 (05:01 -0700)]
Bump aic7xxx and aic79xx driver versions to reflect recent DV changes.

22 years agoAdd a failsafe mechanism to configure devices that have inquiry data
Justin T. Gibbs [Mon, 30 Dec 2002 11:57:48 +0000 (04:57 -0700)]
Add a failsafe mechanism to configure devices that have inquiry data
but somehow are not handled by the DV state machine.  This ensures that
the behavior seen before DV is restored in the event of a DV state machine
failure.

22 years agoClean up check_region() usage. It is deprecated in 2.5.X and 2.4.X,
Justin T. Gibbs [Mon, 30 Dec 2002 11:54:22 +0000 (04:54 -0700)]
Clean up check_region() usage.  It is deprecated in 2.5.X and 2.4.X,
but is still required in earlier kernels.

22 years agoIgnore media not-present errors during DV. This caused DV to fail
Justin T. Gibbs [Mon, 30 Dec 2002 11:52:55 +0000 (04:52 -0700)]
Ignore media not-present errors during DV.  This caused DV to fail
on some tape and cdrom devices.

22 years agoUse down_interruptable() rather than down() to avoid having the DV threads
Justin T. Gibbs [Mon, 30 Dec 2002 11:51:41 +0000 (04:51 -0700)]
Use down_interruptable() rather than down() to avoid having the DV threads
counted toward the load average.

22 years agoFix the last reference to the reg_print.c file handle in symtable_dump.
Justin T. Gibbs [Mon, 30 Dec 2002 11:49:30 +0000 (04:49 -0700)]
Fix the last reference to the reg_print.c file handle in symtable_dump.
This allows the assembler to operate without generating this file.

22 years agoRestore driver style. All functions are declared prior to being defined.
Justin T. Gibbs [Mon, 30 Dec 2002 11:48:20 +0000 (04:48 -0700)]
Restore driver style.  All functions are declared prior to being defined.
The original bug was that the 2.4.X declaration was used unconditionally.

22 years agoCorrectly enable highmem_io option in 2.5.X.
Justin T. Gibbs [Mon, 30 Dec 2002 11:46:51 +0000 (04:46 -0700)]
Correctly enable highmem_io option in 2.5.X.

Daemonize the kernel threads used for domain validation.

22 years agoPreface the "asserting atn" diagnostic with controller/target information.
Justin T. Gibbs [Mon, 30 Dec 2002 11:43:35 +0000 (04:43 -0700)]
Preface the "asserting atn" diagnostic with controller/target information.

Restore a call to ahc_assert_atn() that was inadvertantly lost when the
asserting atn diagnostic was added.  This, among other things, resulted
in driver negotiation failures with devices that initiated negotiation
(e.g. most IBM drives).

22 years agoPreface the "asserting atn" diagnostic with controller/target information.
Justin T. Gibbs [Mon, 30 Dec 2002 11:40:51 +0000 (04:40 -0700)]
Preface the "asserting atn" diagnostic with controller/target information.

Correct a compilation warning that occurs when debugging code is
compiled in.

22 years agoRemove incorrect dependency on SCSI_AIC7XXX_BUILD_FIRMWARE and
Justin T. Gibbs [Mon, 30 Dec 2002 11:38:17 +0000 (04:38 -0700)]
Remove incorrect dependency on SCSI_AIC7XXX_BUILD_FIRMWARE and
AIC7XXX_REG_PRETTY_PRINT.  The constant rebuild is due to the build
process deciding that the pretty_print.c file is an intermediate file
and that problem lies elsewhere.

22 years agoCorrect leading whitespace.
Justin T. Gibbs [Mon, 30 Dec 2002 11:35:45 +0000 (04:35 -0700)]
Correct leading whitespace.

Remove incorrect dependency on SCSI_AIC7XXX_BUILD_FIRMWARE (the
wrong driver noless) and AIC79XX_REG_PRETTY_PRINT.  The constant
rebuild is due to the build process deciding that the pretty_print.c
file is an intermediate file and that problem lies elsewhere.

22 years agoDecend into the aic7xxx directory for the aic79xx driver too.
Justin T. Gibbs [Mon, 30 Dec 2002 11:32:11 +0000 (04:32 -0700)]
Decend into the aic7xxx directory for the aic79xx driver too.

22 years ago[PATCH] more i2c cruft removal
Christoph Hellwig [Mon, 30 Dec 2002 10:07:24 +0000 (02:07 -0800)]
[PATCH] more i2c cruft removal

* version code checks for pre-2.4 version
* !__KERNEL__ stuff in headers.

22 years ago[PATCH] ehci.txt (doc update)
David Brownell [Mon, 30 Dec 2002 09:16:48 +0000 (01:16 -0800)]
[PATCH] ehci.txt (doc update)

Makes this generic info reflect current behavior.

22 years ago[PATCH] cleanup after dead hc needs task context
David Brownell [Mon, 30 Dec 2002 09:16:25 +0000 (01:16 -0800)]
[PATCH] cleanup after dead hc needs task context

Simple patch to invoke hcd->stop() in task context, as
required.  When Cardbus works again (broken in 2.5.53
unless it's just me), this will get rid of some oopsing
when folk physically eject the device, with no shutdown.
As well as making other "hc died" faults behave better.

22 years agoUSB: fix up init_module and cleanup_module mess in speedtouch driver
Greg Kroah-Hartman [Mon, 30 Dec 2002 09:15:20 +0000 (01:15 -0800)]
USB: fix up init_module and cleanup_module mess in speedtouch driver

22 years agoMerge bk://kernel.bkbits.net/gregkh/linux/pnp-2.5
Linus Torvalds [Mon, 30 Dec 2002 07:37:08 +0000 (23:37 -0800)]
Merge bk://kernel.bkbits.net/gregkh/linux/pnp-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoMerge bk://linux-dj.bkbits.net/agpgart
Linus Torvalds [Mon, 30 Dec 2002 07:36:34 +0000 (23:36 -0800)]
Merge bk://linux-dj.bkbits.net/agpgart
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoMerge halogen.(none):/home/davej/src/kernel/2.5/bk-linus
Dave Jones [Mon, 30 Dec 2002 23:08:22 +0000 (23:08 +0000)]
Merge halogen.(none):/home/davej/src/kernel/2.5/bk-linus
into halogen.(none):/home/davej/src/kernel/2.5/agpgart

22 years ago[AGP] propagate agp_backend_acquire() return code up to the ioctl.
Dave Jones [Mon, 30 Dec 2002 22:26:57 +0000 (22:26 +0000)]
[AGP] propagate agp_backend_acquire() return code up to the ioctl.

22 years ago[AGP] set the AMD8151 agp_bridge.type if we detected it.
Dave Jones [Mon, 30 Dec 2002 21:17:05 +0000 (21:17 +0000)]
[AGP] set the AMD8151 agp_bridge.type if we detected it.

22 years ago[AGP] CONFIG_AGP3 depends on CONFIG_AGP
Dave Jones [Mon, 30 Dec 2002 15:59:50 +0000 (14:59 -0100)]
[AGP] CONFIG_AGP3 depends on CONFIG_AGP

22 years agoMerge tetrachloride.(none):/mnt/stuff/kernel/2.5/bk-linus
Dave Jones [Mon, 30 Dec 2002 09:06:22 +0000 (08:06 -0100)]
Merge tetrachloride.(none):/mnt/stuff/kernel/2.5/bk-linus
into tetrachloride.(none):/mnt/stuff/kernel/2.5/agpgart

22 years agofix null agp_bridge.dev with KT400
Leopoldo Cerbaro [Mon, 30 Dec 2002 08:02:38 +0000 (07:02 -0100)]
fix null agp_bridge.dev with KT400

22 years ago[PATCH] eliminate __kernel_..._t32 from s390x
Stephen Rothwell [Mon, 30 Dec 2002 07:35:17 +0000 (23:35 -0800)]
[PATCH] eliminate __kernel_..._t32 from s390x

With Martin's blessing, here is the s390x syscall compatability stuff.

22 years ago[PATCH] fix fs/dquot.c compilation
Christoph Hellwig [Mon, 30 Dec 2002 07:35:11 +0000 (23:35 -0800)]
[PATCH] fix fs/dquot.c compilation

Another one that James noticed..

22 years agoMerge kroah.com:/home/linux/linux/BK/bleeding-2.5
Greg Kroah-Hartman [Mon, 30 Dec 2002 06:17:54 +0000 (22:17 -0800)]
Merge kroah.com:/home/linux/linux/BK/bleeding-2.5
into kroah.com:/home/linux/linux/BK/pnp-2.5

22 years ago[PATCH] Fix hw_irq to test the proper CONFIG variable
James Bottomley [Mon, 30 Dec 2002 06:03:14 +0000 (22:03 -0800)]
[PATCH] Fix hw_irq to test the proper CONFIG variable

22 years agoMerge
Greg Kroah-Hartman [Mon, 30 Dec 2002 05:49:02 +0000 (21:49 -0800)]
Merge

22 years ago[PATCH] Missed one 'try_inc_mod_count()'
Christoph Hellwig [Mon, 30 Dec 2002 05:48:11 +0000 (21:48 -0800)]
[PATCH] Missed one 'try_inc_mod_count()'

I missed one hunk when splitting the patch for removing
try_inc_mod_count().

22 years agoMake x86 platform choice strings more easily selectable
Linus Torvalds [Mon, 30 Dec 2002 05:47:09 +0000 (21:47 -0800)]
Make x86 platform choice strings more easily selectable

22 years ago[PATCH] Don't make the slab might_sleep() check dependent on slab debugging
Andrew Morton [Mon, 30 Dec 2002 05:41:30 +0000 (21:41 -0800)]
[PATCH] Don't make the slab might_sleep() check dependent on slab debugging

The might_sleep() check in kmm_cache_alloc() is currently disabled if slab
debugging is disabled.

Fix that so it's enabled if might_sleep debugging is enabled.

22 years ago[PATCH] Reduce context switch rate due to the random driver
Andrew Morton [Mon, 30 Dec 2002 05:41:23 +0000 (21:41 -0800)]
[PATCH] Reduce context switch rate due to the random driver

add_disk_randomness() is causing a context switch per disk request.  It
is scheduling process-context work one timer tick in the future for
every request.

But is has a buffer for this, so change it to not set up the
process-context work until that buffer is half full.

The patch reduces the context switch rate during a 20 megabyte/sec
write to scsi from 150/sec to 50/sec.

22 years ago[PATCH] don't call console drivers on non-online CPUs
Andrew Morton [Mon, 30 Dec 2002 05:41:16 +0000 (21:41 -0800)]
[PATCH] don't call console drivers on non-online CPUs

George Anzinger identified the following problem: when a secondary CPU is
coming up, it calls printk() before it is "online".  It calls the console
drivers before its per-cpu storage has been prepared.  And the vga console
driver does a mod_timer().  This CPU's timers have not yet been initialised;
it is not clear why this doesn't oops - George thinks it is because virtual
address zero is still accessible at that time.

I believe the right way to fix this is to change printk so that a not-online
CPU will not call the console drivers.  Because printk should always be
callable.  If the CPU is not online the message is buffered, so the next
caller to printk who is online will actually display it.

ia64 has been doing exactly this for ages, so we can remove the
arch_consoles_callable() hook and just open-code the cpu_online() test in
printk.

That fixes things up for the secondary CPUs.  But this change causes a
problem for the boot CPU: it is being marked online very late in boot, so the
printk buffer is being displayed much later than we would like.

I believe that the solution to this is to mark the boot CPU online much
earlier.  So in this patch we call the new arch-provided function
smp_prepare_boot_cpu() immediately after the boot CPU's per-cpu areas are set
up.  Its mandate is to (at least) mark the boot CPU "online".

The change has been reviewed by davem and rth.  No comments were received
from the other arch maintainers.

22 years ago[PATCH] promote the ALIGN() macro
Andrew Morton [Mon, 30 Dec 2002 05:41:09 +0000 (21:41 -0800)]
[PATCH] promote the ALIGN() macro

ALIGN() currently has global scope in <linux/cache.h>.  This causes a
compilation error in the defxx driver.

Move ALIGN() to <linux/kernel.h> and change the defxx driver to use the
generic macro in place of its own.

22 years ago[PATCH] Fix missing brelse() on ext3 htree error path
Andrew Morton [Mon, 30 Dec 2002 05:41:03 +0000 (21:41 -0800)]
[PATCH] Fix missing brelse() on ext3 htree error path

It will leak a bh ref on ENOMEM.

22 years ago[PATCH] remove hugetlb syscalls
Andrew Morton [Mon, 30 Dec 2002 05:40:57 +0000 (21:40 -0800)]
[PATCH] remove hugetlb syscalls

Patch from Ben LaHaise and Bill Irwin.

As discussed a month or so ago, all parties agree that the new
hugetlbfs and shm APIs are sufficient for the 2.6 kernel.

I marked the vacated syscall slots as "available for reuse".  I doubt
that there will be any conflicts when that happens.

22 years ago[PATCH] remove task_struct.swappable
Andrew Morton [Mon, 30 Dec 2002 05:40:51 +0000 (21:40 -0800)]
[PATCH] remove task_struct.swappable

Remove unused task_struct.swappable.

22 years ago[PATCH] Don't cacheline-align vm_area_struct
Andrew Morton [Mon, 30 Dec 2002 05:40:44 +0000 (21:40 -0800)]
[PATCH] Don't cacheline-align vm_area_struct

Some workloads (Oracle...) use a huge number of VMA's.  They are
currently a tidy 64 bytes in size, and padding them out to 128 on P4's
is not worthwhile.

22 years ago[PATCH] kmalloc_percpu -- stripped down version
Andrew Morton [Mon, 30 Dec 2002 05:40:37 +0000 (21:40 -0800)]
[PATCH] kmalloc_percpu -- stripped down version

Patch from Ravikiran G Thirumalai <kiran@in.ibm.com>

Creates a simple "kmalloc for each CPU" API.  This will be used for net
statistics, disk statistics, etc.  (davem has acked the net patches which use
this code).

kmalloc_per_cpu() is available to modules, unlike the current static per-cpu
infrastructure.

22 years ago[PATCH] add drain_local_pages() for CONFIG_SOFTWARE_SUSPEND
Andrew Morton [Mon, 30 Dec 2002 05:40:31 +0000 (21:40 -0800)]
[PATCH] add drain_local_pages() for CONFIG_SOFTWARE_SUSPEND

swsusp gets confused when pages which it freed do not appear in the
buddy lists.  So provide a function which will drain the calling CPU's
per-cpu-pages into the buddy.

The patch has been tested by Pavel.  Presence of the new code is
conditional on CONFIG_SOFTWARE_SUSPEND.

22 years ago[PATCH] Enable semtimedop for ia64 32-bit emulation.
Andrew Morton [Mon, 30 Dec 2002 05:40:25 +0000 (21:40 -0800)]
[PATCH] Enable semtimedop for ia64 32-bit emulation.

Patch from Mark Fasheh <mark.fasheh@oracle.com>

A missing chunk from the semtimedop() implementation.  It enables
semtimedop() for 32-bit ia32 applications running on ia64.

It also changes semop() to go direct to sys_semtimedop(), bypassing a
function call.

22 years ago[PATCH] BIN_TO_BCD consolidation
Andrew Morton [Mon, 30 Dec 2002 05:40:19 +0000 (21:40 -0800)]
[PATCH] BIN_TO_BCD consolidation

Cleanup patch from Hollis Blanchard <hollis@austin.ibm.com>

We have a large number of private implementations of BIN_TO_BCD and
BCD_TO_BIN, which are all the same.  And a lot of them are inflexible because
they modify their arg:

#define BIN_TO_BCD(val) ((val)=(((val)/10)<<4) + (val)%10)

- Create (in <linux/bcd.h> a generic BIN2BCD/BCD2BIN which does not modify
  its arg

- Create generic BIN_TO_BCD/BCD_TO_BIN which uses the above

- Update lots of callers to use the new generic version.

22 years ago[PATCH] Remove /proc/meminfo:MemShared
Andrew Morton [Mon, 30 Dec 2002 05:40:13 +0000 (21:40 -0800)]
[PATCH] Remove /proc/meminfo:MemShared

It seems to have been displaying zero for the past several years, and
we can't think of anything useful to put in there.

22 years ago[PATCH] Docs: fix explanation of file-nr
Andrew Morton [Mon, 30 Dec 2002 05:40:06 +0000 (21:40 -0800)]
[PATCH] Docs: fix explanation of file-nr

Patch from Paul Jakma <paulj@alphyra.ie>

Documentation/sysctl/fs.txt is incorrect wrt to the middle number of
/proc/sys/fs/file-nr.  The current docs state it is the number of
in-use file-handles, from observation this number is actually the
number of /unused/ file-handles - we've observe that when it hits 0 we
get file handle problems (rather than when it hits == file-max as docs
imply).

22 years agoMerge
Greg Kroah-Hartman [Mon, 30 Dec 2002 05:39:30 +0000 (21:39 -0800)]
Merge

22 years agoMerge
Linus Torvalds [Mon, 30 Dec 2002 05:29:39 +0000 (21:29 -0800)]
Merge

22 years ago[PATCH] MODULE_PARM "c" support
Rusty Russell [Mon, 30 Dec 2002 05:04:02 +0000 (21:04 -0800)]
[PATCH] MODULE_PARM "c" support

Turns out there was an undocumented "c" flag for MODULE_PARM.

This implementation is a little ugly, but it works, and will do for
compatibility (I haven't implemented such a two-dimensional array
primitive, but the whole point of the module_parm et al is that they are
extensible).

22 years ago[PATCH] rename CONFIG_VOYAGER to CONFIG_X86_VOYAGER
Christoph Hellwig [Mon, 30 Dec 2002 04:57:35 +0000 (20:57 -0800)]
[PATCH] rename CONFIG_VOYAGER to CONFIG_X86_VOYAGER

The name is just a bit too generic, and we already use the _X86 prefix
for lots of other stuff in that area.  Dito for the never use CONFIG_PC
and CONFIG_VISWS.

22 years ago[PATCH] crc32 speedup/use anywhere
Brian Murphy [Mon, 30 Dec 2002 04:33:32 +0000 (20:33 -0800)]
[PATCH] crc32 speedup/use anywhere

This patch combines my patch which statically initialises the crc32
tables so they can be used at any time (during initialisation) and
Joakim Tjernlund's patch to speed up the crc calculations by doing word
operations instead of exclusively byte.

The crc routines are used extensively in jffs2 where speed is very
important.

I need the crc32 routines to calculate a checksum on values read from an
eeprom which contain cpu speed and memory size information - so they are
needed very much earlier in the initialisation process than they are
currently available.

22 years ago[PATCH] cpufreq: remove usage of #typedef
Dominik Brodowski [Mon, 30 Dec 2002 04:29:35 +0000 (20:29 -0800)]
[PATCH] cpufreq: remove usage of #typedef