]> git.hungrycats.org Git - linux/log
linux
22 years agoLinux v2.5.57 v2.5.57
Linus Torvalds [Mon, 13 Jan 2003 02:11:49 +0000 (18:11 -0800)]
Linux v2.5.57

22 years ago[PATCH] P4-based Celeron comments
Robert Love [Mon, 13 Jan 2003 01:38:50 +0000 (17:38 -0800)]
[PATCH] P4-based Celeron comments

As you mentioned, we do not correctly identify the P4-based Celeron in
the kernel configuration help.  Unfortunately, Intel has called all
Celeron products simply "Celeron", so we call these "P4-based Celerons".

22 years agoGet rid of endless loop in PnP-enabled ne.c
Linus Torvalds [Mon, 13 Jan 2003 00:42:20 +0000 (16:42 -0800)]
Get rid of endless loop in PnP-enabled ne.c

22 years ago[PATCH] Fix PnP BIOS fault handling
Brian Gerst [Mon, 13 Jan 2003 00:36:55 +0000 (16:36 -0800)]
[PATCH] Fix PnP BIOS fault handling

Check for PnP BIOS in all fault paths, not just in do_trap().

22 years agoMerge http://ppc.bkbits.net/for-linus-ppc64
Linus Torvalds [Mon, 13 Jan 2003 00:28:27 +0000 (16:28 -0800)]
Merge http://ppc.bkbits.net/for-linus-ppc64
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoppc64: compat_sys_[f]statfs, from Stephen Rothwell
Anton Blanchard [Tue, 14 Jan 2003 05:35:24 +0000 (16:35 +1100)]
ppc64: compat_sys_[f]statfs, from Stephen Rothwell

22 years agoMerge samba.org:/scratch/anton/for-alan
Anton Blanchard [Tue, 14 Jan 2003 05:22:18 +0000 (16:22 +1100)]
Merge samba.org:/scratch/anton/for-alan
into samba.org:/scratch/anton/sfr

22 years agoMerge samba.org:/scratch/anton/linux-2.5
Anton Blanchard [Tue, 14 Jan 2003 05:19:13 +0000 (16:19 +1100)]
Merge samba.org:/scratch/anton/linux-2.5
into samba.org:/scratch/anton/sfr

22 years agoppc64: zero extend all 6 parameters in 32 bit syscall path
Anton Blanchard [Tue, 14 Jan 2003 04:51:59 +0000 (15:51 +1100)]
ppc64: zero extend all 6 parameters in 32 bit syscall path

22 years agoMerge samba.org:/scratch/anton/linux-2.5
Anton Blanchard [Tue, 14 Jan 2003 04:39:43 +0000 (15:39 +1100)]
Merge samba.org:/scratch/anton/linux-2.5
into samba.org:/scratch/anton/for-alan

22 years agoMerge samba.org:/scratch/anton/linux-2.5
Anton Blanchard [Tue, 14 Jan 2003 01:21:33 +0000 (12:21 +1100)]
Merge samba.org:/scratch/anton/linux-2.5
into samba.org:/scratch/anton/for-alan

22 years agoppc64: remove mmu_gather_t
Anton Blanchard [Mon, 13 Jan 2003 05:45:50 +0000 (16:45 +1100)]
ppc64: remove mmu_gather_t

22 years agoMerge samba.org:/scratch/anton/linux-2.5
Anton Blanchard [Mon, 13 Jan 2003 05:10:27 +0000 (16:10 +1100)]
Merge samba.org:/scratch/anton/linux-2.5
into samba.org:/scratch/anton/for-alan

22 years agoMerge bk://kernel.bkbits.net/davem/net-2.5
Linus Torvalds [Mon, 13 Jan 2003 00:28:01 +0000 (16:28 -0800)]
Merge bk://kernel.bkbits.net/davem/net-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoMerge bk://kernel.bkbits.net/davem/sparc-2.5
Linus Torvalds [Mon, 13 Jan 2003 00:27:32 +0000 (16:27 -0800)]
Merge bk://kernel.bkbits.net/davem/sparc-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoMerge bk://ppc.bkbits.net/for-linus-ppc
Linus Torvalds [Mon, 13 Jan 2003 00:27:08 +0000 (16:27 -0800)]
Merge bk://ppc.bkbits.net/for-linus-ppc
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoPPC32: Change struct free_pte_ctx to struct mmu_gather.
Paul Mackerras [Tue, 14 Jan 2003 09:22:38 +0000 (20:22 +1100)]
PPC32: Change struct free_pte_ctx to struct mmu_gather.

A couple of occurrences of struct free_pte_ctx in include/asm-ppc/tlb.h
got missed in akpm's patch - this fixes them.

22 years agoPPC32: Fix copy_from_user to copy as much as possible.
Dale Farnsworth [Tue, 14 Jan 2003 09:17:16 +0000 (20:17 +1100)]
PPC32: Fix copy_from_user to copy as much as possible.

copy_from_user is supposed to transfer as much data as is
valid and then to return the number of bytes not tranferred.
That's how it works on x86.  On ppc it can be as much as 15
bytes short.

I initially saw the problem with the mount system call.

Note that the fifth argument to mount is an address 8 bytes from the end
of user data space.  There is a null byte at that address, since no mount
options are being passed.

In the kernel, sys_mount() allocates a page for the options and
does copy_from_user(new_page, 0x1005eff8, PAGE_SIZE).  copy_from_user
should copy 8 bytes and return (PAGE_SIZE-8).  Instead, on ppc it reads
8 bytes, faults, writes no bytes, and returns PAGE_SIZE, which causes the
EFAULT to be erroneously reported.

22 years agoMerge samba.org:/home/paulus/kernel/linux-2.5
Paul Mackerras [Mon, 13 Jan 2003 22:18:16 +0000 (09:18 +1100)]
Merge samba.org:/home/paulus/kernel/linux-2.5
into samba.org:/home/paulus/kernel/for-linus-ppc

22 years ago[COMPAT]: fs/compat.c needs linux/vfs.h for asm/statfs.h
David S. Miller [Sun, 12 Jan 2003 17:16:36 +0000 (09:16 -0800)]
[COMPAT]: fs/compat.c needs linux/vfs.h for asm/statfs.h

22 years ago[SUNZILOG]: Fix spinlock access in previous changes.
David S. Miller [Sun, 12 Jan 2003 17:14:34 +0000 (09:14 -0800)]
[SUNZILOG]: Fix spinlock access in previous changes.

22 years ago[SUNZILOG]: Get serial console et al. working once more.
Pete Zaitcev [Sun, 12 Jan 2003 16:22:10 +0000 (08:22 -0800)]
[SUNZILOG]: Get serial console et al. working once more.
- Make sure R9 is really loaded
- Make sure IS_CONS flag is handled properly.

22 years ago[IPV4]: Allow route.c to build without procfs enabled.
Paul Rolland [Sun, 12 Jan 2003 16:19:02 +0000 (08:19 -0800)]
[IPV4]: Allow route.c to build without procfs enabled.

22 years ago[ebtables] use Rustys new module scheme in ebtables.c, vs 2.5.56
Bart De Schuymer [Sun, 12 Jan 2003 16:16:44 +0000 (08:16 -0800)]
[ebtables] use Rustys new module scheme in ebtables.c, vs 2.5.56
- use /* */ instead of //
- change my mail alias
- remove MOD_DEC_USE_COUNT and MOD_INC_USE_COUNT
- use try_module_get (and check result) and module_put instead of
  __MOD_INC_USE_COUNT and __MOD_DEC_USE_COUNT
- add \n in init/exit messages

22 years ago[TCP]: Dont tcp_listen_unlock unless it was locked.
Anders Gustafsson [Sun, 12 Jan 2003 16:13:38 +0000 (08:13 -0800)]
[TCP]: Dont tcp_listen_unlock unless it was locked.

22 years ago[COMPAT] compat_sys_[f]statfs - sparc64 part
Stephen Rothwell [Sun, 12 Jan 2003 16:10:31 +0000 (08:10 -0800)]
[COMPAT] compat_sys_[f]statfs - sparc64 part

22 years ago[PATCH] compat_sys_[f]statfs - s390x part
Stephen Rothwell [Sun, 12 Jan 2003 13:31:54 +0000 (05:31 -0800)]
[PATCH] compat_sys_[f]statfs - s390x part

Here is the s390x part. Others will go through the respective maintainers.

22 years ago[PATCH] compat_sys_[f]statfs - generic part
Stephen Rothwell [Sun, 12 Jan 2003 13:31:48 +0000 (05:31 -0800)]
[PATCH] compat_sys_[f]statfs - generic part

This patch creates compat_sys_statfs and compat_sys_fstatfs. This is just
the generic part of the patch. Specific archs will follow.

22 years ago[PATCH] v850 obsolete params fix
Rusty Russell [Sun, 12 Jan 2003 12:30:19 +0000 (04:30 -0800)]
[PATCH] v850 obsolete params fix

Since these are just symbols in the module object, they need symbol name
munging to find the symbol from the parameter name.

22 years ago[PATCH] Fix strlen_user usage in module.c
Rusty Russell [Sun, 12 Jan 2003 12:30:13 +0000 (04:30 -0800)]
[PATCH] Fix strlen_user usage in module.c

strlen_user returns 0 on error, not an error number, and otherwise
returns the length including the NUL byte.  Found by Andi Kleen.

22 years ago[PATCH] Remove dup __gpl_ksymtab in arm file
Rusty Russell [Sun, 12 Jan 2003 12:27:35 +0000 (04:27 -0800)]
[PATCH] Remove dup __gpl_ksymtab in arm file

Russell's patch beat mine in, and unfortunately didn't conflict, so
armv has two __gpl_ksymtab sections.

Revert mine.

22 years ago[PATCH] PTRACE_GET_THREAD_AREA
Roland McGrath [Sun, 12 Jan 2003 12:02:53 +0000 (04:02 -0800)]
[PATCH] PTRACE_GET_THREAD_AREA

Add ptrace support for getting and setting the THREAD_AREA of the
thread being debugged on x86.

22 years agoAlways assign bus numbers for cardbus. Firmware often doesn't do it right.
Linus Torvalds [Sun, 12 Jan 2003 11:44:30 +0000 (03:44 -0800)]
Always assign bus numbers for cardbus. Firmware often doesn't do it right.

22 years ago[PATCH] Place __gpl_ksymtab section in all linker scripts
Rusty Russell [Sun, 12 Jan 2003 11:22:00 +0000 (03:22 -0800)]
[PATCH] Place __gpl_ksymtab section in all linker scripts

Explicitly place the __gpl_ksymtab section for every arch.  RMK
pointed out that some archs will place it really badly otherwise.

22 years agoISDN/HiSax: Fix some of the new PnP stuff
Kai Germaschewski [Sun, 12 Jan 2003 13:00:03 +0000 (07:00 -0600)]
ISDN/HiSax: Fix some of the new PnP stuff

I appreciate the adaption of the ISDN drivers to the new PnP layer
(though I don't know why it was actually necessary to break the old
ISAPnP so late in the cycle), but the gratuitious changes to the coding
style were not exactly necessary. So revert things to the
"if (retval) goto err_unwind;" style and fix a couple of other compile
time errors.

22 years agoMerge tp1.ruhr-uni-bochum.de:/scratch/kai/kernel/v2.5/linux-2.5
Kai Germaschewski [Sun, 12 Jan 2003 12:15:41 +0000 (06:15 -0600)]
Merge tp1.ruhr-uni-bochum.de:/scratch/kai/kernel/v2.5/linux-2.5
into tp1.ruhr-uni-bochum.de:/scratch/kai/kernel/v2.5/linux-2.5.isdn

22 years agoISDN/HiSax: Move chipset init code into *_setup() functions
Kai Germaschewski [Sun, 12 Jan 2003 12:12:42 +0000 (06:12 -0600)]
ISDN/HiSax: Move chipset init code into *_setup() functions

This is just again moving some common code into shared places,
let the chip-specific driver check the versions instead of having
every card-specific driver do so.

22 years agoISDN/HiSax: Remove superfluous card type checks
Kai Germaschewski [Sun, 12 Jan 2003 12:05:40 +0000 (06:05 -0600)]
ISDN/HiSax: Remove superfluous card type checks

We use card->typ to decide which subdriver to call, so there's
no need at all for the subdriver to check again that card->typ is correct.

22 years agoISDN/HiSax: Unify LED handling
Kai Germaschewski [Sun, 12 Jan 2003 12:03:34 +0000 (06:03 -0600)]
ISDN/HiSax: Unify LED handling

Share the code to display line status via LEDs and get rid of the
remaining ->cardmsg() users.

22 years agoISDN/HiSax: Remove empty ->cardmsg
Kai Germaschewski [Sun, 12 Jan 2003 12:02:03 +0000 (06:02 -0600)]
ISDN/HiSax: Remove empty ->cardmsg

A lot of the subdrivers don't do anything in their ->cardmsg function
anymore, so we just remove those and make the callers check for NULL.

Also, add ops->bc_{,de}activate() for the drivers which want to be notified
of B-Channel activation.

22 years agoISDN/HiSax: Remove superfluous #ifdef CONFIG_PCI
Kai Germaschewski [Sun, 12 Jan 2003 11:57:44 +0000 (05:57 -0600)]
ISDN/HiSax: Remove superfluous #ifdef CONFIG_PCI

Subdrivers which only drive PCI cards can have the dependency on
CONFIG_PCI moved to Kconfig, removing some ugly #ifdefs.

22 years agoISDN/HiSax: Convert remaining ioremap() to request_mmio()
Kai Germaschewski [Sun, 12 Jan 2003 11:55:53 +0000 (05:55 -0600)]
ISDN/HiSax: Convert remaining ioremap() to request_mmio()

These users didn't bother doing request_mem_region() at all before
ioremapping, which is now automatically done using request_mmio()
instead.

22 years agoISDN/HiSax: Simplified request_region() etc.
Kai Germaschewski [Sun, 12 Jan 2003 11:53:28 +0000 (05:53 -0600)]
ISDN/HiSax: Simplified request_region() etc.

This patch introduces, private to the HiSax driver, new helper functions
request_io/mmio(), which correspond to request_region()/
request_mem_region() but also are verbose about failures and keep track
of the allocated regions, so unwinding in case of errors is automatic.

Additionally, request_mmio() will also ioremap() the region.

22 years agoMerge
Linus Torvalds [Sun, 12 Jan 2003 03:50:08 +0000 (19:50 -0800)]
Merge

22 years ago[PATCH] PnP update - drivers
Jaroslav Kysela [Sun, 12 Jan 2003 03:46:40 +0000 (19:46 -0800)]
[PATCH] PnP update - drivers

this patch must be applied after PnP patch v0.94. It contains my
small cleanups of PnP code and I tried to rewrite almost all ISA PnP
drivers to new PnP subsystem except sound drivers (ALSA & OSS).

  PnP update
    - separated dev->resource to dev->io_resource and dev->mem_resource
    - added pnp_*_valid() macros
    - added status member to pnp_card structure
    - added pnp_device_attach/detach functions
    - added pnpc_attach/detach functions
    - moved pnp_find_card() and pnp_find_dev() functions to isapnp.h
    - updated IDE/gameport/ISDN/network/scsi/radio/telephone drivers
      to latest PnP code

22 years ago[PATCH] 2.5.56, ne2k compiles and works
Martin H. VanLeeuwen [Sun, 12 Jan 2003 03:43:08 +0000 (19:43 -0800)]
[PATCH] 2.5.56, ne2k compiles and works

This patch makes the ne.c compile and ISA pnp ne2k work after
changes in 2.5.54 for PNP broke it.

 a. use pnp_dev instead of pci_dev
 b. convert to appropriate pnp_*()'s
 c. check -1 instead of 0 for invalid IRQ

22 years ago[PATCH] 2.5.56, isapnp cards not found
Martin H. VanLeeuwen [Sun, 12 Jan 2003 03:43:03 +0000 (19:43 -0800)]
[PATCH] 2.5.56, isapnp cards not found

This patch allows isapnp cards to be detected now.  Seems around 2.5.54,
a change in isapnp_build_device_list changed:

card->protocol = &isapnp_protocol;
 to
card->protocol = &isapnp_card_protocol;

This patch adapts isapnp_init to the above change.

22 years ago[PATCH] 2.5.56, isapnp_init level
Martin H. VanLeeuwen [Sun, 12 Jan 2003 03:42:57 +0000 (19:42 -0800)]
[PATCH] 2.5.56, isapnp_init level

In 2.5.50 the level of isapnp_init was moved to after apci.  Since it is
now after net_dev_init, ISA PNP NICS fail to initialized at boot.

This fix allows ISA PNP NIC cards to work during net_dev_init, and still
leaves isapnp_init after apci_init.

22 years agoAutomerge
Linus Torvalds [Sun, 12 Jan 2003 03:42:38 +0000 (19:42 -0800)]
Automerge

22 years ago[PATCH] add explicit Pentium II support
Robert Love [Sun, 12 Jan 2003 03:23:34 +0000 (19:23 -0800)]
[PATCH] add explicit Pentium II support

This separates the "PPro/Celeron/Pentium-II" compile option into "PPro"
and "Pentium-II/Celeron" options.

This allows us to explicitly support the Pentium II and Celeron,
specifically adding the `-march' option for the chip and enabling the
unaligned copy optimizations.  PPro support remains unchanged.

This patch is by Luuk van der Duim with some changes by me (primarily to
also support the pre-Coppermine Celeron chips, since those use Pentium
II cores).  This patch has been in 2.5-mm for awhile and Andrew ack'ed
this submission.

22 years agoMerge bk://bk.arm.linux.org.uk
Linus Torvalds [Sun, 12 Jan 2003 03:22:24 +0000 (19:22 -0800)]
Merge bk://bk.arm.linux.org.uk
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years ago[ARM] Bring sa1100_ir.c up to date wrt new DMA and device subsystems.
Russell King [Sun, 12 Jan 2003 17:16:37 +0000 (17:16 +0000)]
[ARM] Bring sa1100_ir.c up to date wrt new DMA and device subsystems.

Update sa1100_ir to use new dma-mapping subsystem.  Register a
system device for the SIR/FIR uart port, and a device driver for
this driver.

22 years ago[ARm] Fix ARM exception table fixups for 2.5.55 updates.
Russell King [Sun, 12 Jan 2003 17:11:52 +0000 (17:11 +0000)]
[ARm] Fix ARM exception table fixups for 2.5.55 updates.

22 years ago[ARM] Fix Jornada720 sa1100-flash.c support, update to C99 initialisers
Russell King [Sun, 12 Jan 2003 17:09:10 +0000 (17:09 +0000)]
[ARM] Fix Jornada720 sa1100-flash.c support, update to C99 initialisers

22 years ago[ARM] Remove set_irq_type, sa1111 driver names are lower case.
Russell King [Sun, 12 Jan 2003 16:47:48 +0000 (16:47 +0000)]
[ARM] Remove set_irq_type, sa1111 driver names are lower case.

set_irq_type is redundant for many SA1111 drivers.  The active edge
is setup by the core SA1111 code for the common case.

SA1111 driver names are lower case.  Make the sa1111 ohci driver
reflect this.

22 years ago[ARM] Make CONFIG_SERIO_RPCKBD default to y for Acorn platforms.
Russell King [Sun, 12 Jan 2003 16:43:42 +0000 (16:43 +0000)]
[ARM] Make CONFIG_SERIO_RPCKBD default to y for Acorn platforms.

22 years ago[ARM] Update extable.c for 2.5.55 exception table / module changes.
Russell King [Sun, 12 Jan 2003 16:32:13 +0000 (16:32 +0000)]
[ARM] Update extable.c for 2.5.55 exception table / module changes.

22 years ago[ARM] Add support for the StrongARM-11x0 watchdog.
Russell King [Sun, 12 Jan 2003 15:26:53 +0000 (15:26 +0000)]
[ARM] Add support for the StrongARM-11x0 watchdog.

Add sa1100_wdt watchdog support.

22 years ago[ARM] Ensure that dev->dma_mask is initialised for Acorn cards.
Russell King [Sun, 12 Jan 2003 14:50:41 +0000 (14:50 +0000)]
[ARM] Ensure that dev->dma_mask is initialised for Acorn cards.

22 years ago[ARM] Update DMA usage in Acorn SCSI drivers
Russell King [Sun, 12 Jan 2003 14:44:00 +0000 (14:44 +0000)]
[ARM] Update DMA usage in Acorn SCSI drivers

Remove the pci_dma functionality from the Acorn SCSI drivers, and
replace with the new DMA API, passing in the generic struct device.

22 years ago[ARM] Update ARM documentation
Russell King [Sun, 12 Jan 2003 12:07:54 +0000 (12:07 +0000)]
[ARM] Update ARM documentation

Add 00-INDEX and booting documentation, remove obsolete DMA
documentation.

22 years ago[ARM] Invalidate TLB before and after setting up page tables
Russell King [Sun, 12 Jan 2003 10:41:45 +0000 (10:41 +0000)]
[ARM] Invalidate TLB before and after setting up page tables

22 years ago[ARM] Add __gpl_ksymtab section to linker script
Russell King [Sun, 12 Jan 2003 10:16:34 +0000 (10:16 +0000)]
[ARM] Add __gpl_ksymtab section to linker script

22 years ago[PATCH] Update 2.5 PIIX driver to match 2.4 PIIX
Alan Cox [Sun, 12 Jan 2003 03:19:39 +0000 (19:19 -0800)]
[PATCH] Update 2.5 PIIX driver to match 2.4 PIIX

This adds knowledge of the relevant Intel errata

- 450NX rev B0 we must not do IDE DMA
- 450NX rev C0 we must ensure restreaming is off or risk corruption
- ICH0-ICH2 have a data corrupting hardware bug in the old old MWDMA0
mode, so we avoid MWDMA0 on such devices

22 years ago[PATCH] Add compaq triflex IDE from 2.4.21pre into 2.5
Alan Cox [Sun, 12 Jan 2003 03:19:31 +0000 (19:19 -0800)]
[PATCH] Add compaq triflex IDE from 2.4.21pre into 2.5

Author:  Torben Mathiasen

Driver for Compaq Triflex IDE - older bus mastering IDE found on Compaq
WS5000 and similar systems.

22 years agoFix typo in the network driver pad fixes for a2065 driver.
Linus Torvalds [Sun, 12 Jan 2003 02:59:48 +0000 (18:59 -0800)]
Fix typo in the network driver pad fixes for a2065 driver.

Noted by Maciej Rozycki <macro@ds2.pg.gda.pl>.

22 years agoModern gcc's don't like labels without a statement.
Linus Torvalds [Sun, 12 Jan 2003 00:52:18 +0000 (16:52 -0800)]
Modern gcc's don't like labels without a statement.

Add empty statements to avoid warning.

22 years agoMake MOD_[INC|DEC]_USE_COUNT a true no-op for built-in modules.
Linus Torvalds [Sun, 12 Jan 2003 00:51:22 +0000 (16:51 -0800)]
Make MOD_[INC|DEC]_USE_COUNT a true no-op for built-in modules.

This also avoids the nasty warnings that nobody seems to bother
to fix.

22 years agoRe-instate the SA_RESTORER functionality, since it seems that some
Linus Torvalds [Sun, 12 Jan 2003 00:50:14 +0000 (16:50 -0800)]
Re-instate the SA_RESTORER functionality, since it seems that some
programs still depend on it and in fact do install a different signal
restorer than the standard kernel version.

22 years ago[PATCH] x86 page table initialization fix
Brian Gerst [Sun, 12 Jan 2003 00:48:49 +0000 (16:48 -0800)]
[PATCH] x86 page table initialization fix

At bootup, one_page_table_init() pulls the rug out from under the kernel
by installing a new page table before setting it up.  With big TLB's, it
can go unnoticed, but a 486 has a small TLB so any miss will cause a
triple fault and reset.

22 years ago[ARM] CPUFREQ updates from Dominik
Russell King [Sat, 11 Jan 2003 23:12:02 +0000 (23:12 +0000)]
[ARM] CPUFREQ updates from Dominik

cpu-sa11x0: CPUFREQ_ALL_CPUS shouldn't be passed as argument to
cpufreq_notify_transition() any more, but the exact CPU number.

mach-integrator: cpufreq_notify_transition() isn't called -- which
means that the transition notifiers are not informed. This patch adds
these callbacks. Also, CPUFREQ_ALL_CPUS can't passed to
cpufreq_driver->set_policy any longer, so some code can safely be
removed.

22 years agoLinux PnP Support 0.94
Jaroslav Kysela [Sat, 11 Jan 2003 11:57:09 +0000 (12:57 +0100)]
Linux PnP Support 0.94

(Adam Belay)            -use list_del instead of list_del_init in some areas
                        -introduce pnp capability and status flags
                        -remove static resource setting, I did some research and found that only
                         PnPBIOS supports it, therefore it is better to implememt this in the
                         PnPBIOS protocol itself. (it appears ACPI doesn't use this)
                        -Remove pnp_dev_has_driver and use PNP_ATTACHED instead, this is necessary
                         because a card driver only has rights over a device that it requests.
                        -added card_for_each_dev macro
                        -undo isapnp protocol changes, the pnp layer was designed to handle cards
                         and devices on the same protocol and I feel they should not be seperated.
(Pual Laufer)           -Fix remove driver bug in pnp card services
(Adam Richter)          -Fix a potential oops in id registration functions

22 years agoMerge http://linux-isdn.bkbits.net/linux-2.5.isdn
Linus Torvalds [Sat, 11 Jan 2003 04:47:51 +0000 (20:47 -0800)]
Merge http://linux-isdn.bkbits.net/linux-2.5.isdn
into home.transmeta.com:/home/torvalds/v2.5/linux

22 years agoMerge tp1.ruhr-uni-bochum.de:/scratch/kai/kernel/v2.5/linux-2.5
Kai Germaschewski [Sat, 11 Jan 2003 08:10:26 +0000 (02:10 -0600)]
Merge tp1.ruhr-uni-bochum.de:/scratch/kai/kernel/v2.5/linux-2.5
into tp1.ruhr-uni-bochum.de:/scratch/kai/kernel/v2.5/linux-2.5.isdn

22 years agoISDN: remove kernel 2.0 code
Kai Germaschewski [Sat, 11 Jan 2003 07:59:10 +0000 (01:59 -0600)]
ISDN: remove kernel 2.0 code

From: Adrian Bunk <bunk@fs.tum.de>

The patch below removes #if'd kernel 2.0 code from
drivers/isdn/divert/divert_init.c.

22 years agoISDN: isdn-tty driver not HZ aware
Kai Germaschewski [Sat, 11 Jan 2003 07:57:46 +0000 (01:57 -0600)]
ISDN: isdn-tty driver not HZ aware

From: Christian Borntraeger <linux@borntraeger.net>

This patch makes isdn_tty HZ aware.
The first change changes 3000 jiffies (now 3 seconds) to 30 seconds according to
the comment.
I dont know, if the second change (schedule_timeout(50);) has to be half a
second but this was the value used in 2.4.

22 years agoISDN/HiSax: Add missing __devexit_p()
Kai Germaschewski [Sat, 11 Jan 2003 07:57:00 +0000 (01:57 -0600)]
ISDN/HiSax: Add missing __devexit_p()

22 years agoISDN/HiSax: Clean up the gazel subdriver
Kai Germaschewski [Sat, 11 Jan 2003 07:56:05 +0000 (01:56 -0600)]
ISDN/HiSax: Clean up the gazel subdriver

Instead of having "switch (subtype)" in just about every function,
rather use separate functions and invoke the right one using
the now existing struct card_ops infrastructure.

22 years agoISDN/HiSax: Share IPAC IRQ handler
Kai Germaschewski [Sat, 11 Jan 2003 07:55:24 +0000 (01:55 -0600)]
ISDN/HiSax: Share IPAC IRQ handler

All IRQ handlers for IPAC based cards were basically the same (not
a big surprise, since the chip is the same), so we can share
the IRQ handler.

22 years agoISDN/HiSax: Generate D/B channel access functions for IPAC
Kai Germaschewski [Sat, 11 Jan 2003 07:43:22 +0000 (01:43 -0600)]
ISDN/HiSax: Generate D/B channel access functions for IPAC

IPAC is basically a combined HSCX/ISAC chip, so we can generate
the D- and B-channel access functions knowing how to access the IPAC.

For performance reasons, this happens in a macro.

22 years agoISDN/HiSax: Clean up the various IPAC IRQ handlers
Kai Germaschewski [Sat, 11 Jan 2003 07:40:27 +0000 (01:40 -0600)]
ISDN/HiSax: Clean up the various IPAC IRQ handlers

Just renaming and introducing some helpers makes them look very similar
to each other..

22 years agoISDN/HiSax: Share interrupt handler for ISAC/HSCX cards
Kai Germaschewski [Sat, 11 Jan 2003 07:37:57 +0000 (01:37 -0600)]
ISDN/HiSax: Share interrupt handler for ISAC/HSCX cards

Except for a minor performance penalty, using the same IRQ handler
for cards which used the same code anyway seems perfectly natural...

22 years agoISDN/HiSax: Share some common D-channel init code
Kai Germaschewski [Sat, 11 Jan 2003 07:24:13 +0000 (01:24 -0600)]
ISDN/HiSax: Share some common D-channel init code

Again, just killing some duplicated code.

22 years agoISDN/HiSax: Move open/close of D-channel stack -> dc_l1_ops
Kai Germaschewski [Sat, 11 Jan 2003 07:18:23 +0000 (01:18 -0600)]
ISDN/HiSax: Move open/close of D-channel stack -> dc_l1_ops

Same change which happened for the B-channel earlier.

22 years agoISDN/HiSax: Introduce methods for reset/test/release/
Kai Germaschewski [Sat, 11 Jan 2003 07:13:18 +0000 (01:13 -0600)]
ISDN/HiSax: Introduce methods for reset/test/release/

This mostly finishes splitting up the multiplexing ->cardmsg.

22 years agoISDN/HiSax: Move interrupt function to per-card struct
Kai Germaschewski [Sat, 11 Jan 2003 07:07:51 +0000 (01:07 -0600)]
ISDN/HiSax: Move interrupt function to per-card struct

Since we now have a per-card ops struct, use it to provide the
irq handler function, too.

Some drivers actually drive more than one specific hardware card,
instead of having "switch (cs->subtyp)" scattered around, we rather
aim at having different card_ops structures which just provide the
right functions for the hardware. Of course, this patch is only the
beginning of that separation, but allows for some cleaning already.

22 years agoISDN/HiSax: Introduce per-card init function
Kai Germaschewski [Sat, 11 Jan 2003 06:56:25 +0000 (00:56 -0600)]
ISDN/HiSax: Introduce per-card init function

Linux normally uses separate callbacks instead of a multiplexing
function like "cardmsg". So start to break that into pieces.

22 years agoMerge uidc2-166.inav.uiowa.net:kernel/v2.5/linux-2.5.isdn
Kai Germaschewski [Sat, 11 Jan 2003 04:36:01 +0000 (22:36 -0600)]
Merge uidc2-166.inav.uiowa.net:kernel/v2.5/linux-2.5.isdn
into tp1.ruhr-uni-bochum.de:/scratch/kai/kernel/v2.5/linux-2.5.isdn

22 years agoPPC32: Use timing workaround for DS1743 RTC chip.
Matt Porter [Sat, 11 Jan 2003 02:50:28 +0000 (13:50 +1100)]
PPC32: Use timing workaround for DS1743 RTC chip.

22 years ago[PATCH] inline 1,2 and 4-byte copy_*_user operations
Andrew Morton [Sat, 11 Jan 2003 02:40:35 +0000 (18:40 -0800)]
[PATCH] inline 1,2 and 4-byte copy_*_user operations

The patch arranges for constant 1, 2 and 4-byte copy_*_user() invokations to
be inlined.

It's hard to tell really, but the AIM9 creat_clo, signal_test and dir_rtns_1
numbers went up by 3%-9%, which is to be expected.

22 years ago[PATCH] fix set_page_dirty vs truncate&free races
Andrew Morton [Sat, 11 Jan 2003 02:40:27 +0000 (18:40 -0800)]
[PATCH] fix set_page_dirty vs truncate&free races

set_page_dirty() is racy if the caller has no reference against
page->mapping->host, and if the page is unlocked.  This is because
another CPU could truncate the page off the mapping and then free the
mapping.

Usually, the page _is_ locked, or the caller is a user-space process which
holds a reference on the inode by having an open file.

The exceptional cases are where the page was obtained via
get_user_pages().  The patch changes those to lock the page around the
set_page_dirty() call.

22 years ago[PATCH] misc fixes
Andrew Morton [Sat, 11 Jan 2003 02:40:20 +0000 (18:40 -0800)]
[PATCH] misc fixes

- Fix error-path mem leak in __vfs_follow_link() (From a recent AC->2.4
  patch)

- Make drivers/net/aironet4500_proc.c:driver_lock static.

22 years ago[PATCH] Fix an SMP+preempt latency problem
Andrew Morton [Sat, 11 Jan 2003 02:40:13 +0000 (18:40 -0800)]
[PATCH] Fix an SMP+preempt latency problem

Here is spin_lock():

#define spin_lock(lock) \
do { \
        preempt_disable(); \
        _raw_spin_lock(lock); \
} while(0)

Here is the scenario:

CPU0:
spin_lock(some_lock);
do_very_long_thing(); /* This has cond_resched()s in it */

CPU1:
spin_lock(some_lock);

Now suppose that the scheduler tries to schedule a task on CPU1.  Nothing
happens, because CPU1 is spinning on the lock with preemption disabled.  CPU0
will happliy hold the lock for a long time because nobody has set
need_resched() against CPU0.

This problem can cause scheduling latencies of many tens of milliseconds on
SMP on kernels which handle UP quite happily.

This patch fixes the problem by changing the spin_lock() and write_lock()
contended slowpath to spin on the lock by hand, while polling for preemption
requests.

I would have done read_lock() too, but we don't seem to have read_trylock()
primitives.

The patch also shrinks the kernel by 30k due to not having separate
out-of-line spinning code for each spin_lock() callsite.

22 years ago[PATCH] low-latency pagetable teardown
Andrew Morton [Sat, 11 Jan 2003 02:40:06 +0000 (18:40 -0800)]
[PATCH] low-latency pagetable teardown

Pagetable teardown can hold page_table_lock for extremely long periods -
hundreds of milliseconds.  This is pretty much the final source of high
scheduling latency in the core kernel.

We fixed it for zap_page_range() by chunking the work up and dropping the
lock occasionally if needed.  But that did not fix exit_mmap() and
unmap_region().

So what this patch does is to create an uber-zapper "unmap_vmas()" which
provides all the vma-walking, page unmapping and low-latency lock-dropping
which zap_page_range(), exit_mmap() and unmap_region() require.  Those three
functions are updated to call unmap_vmas().

It's actually a bit of a cleanup...

22 years ago[PATCH] Don't reverse the VMA list in touched_by_munmap()
Andrew Morton [Sat, 11 Jan 2003 02:39:59 +0000 (18:39 -0800)]
[PATCH] Don't reverse the VMA list in touched_by_munmap()

touched_by_munmap() returns a reversed list of VMA's.  That makes things
harder in the low-latency-page-zapping patch.

So change touched_by_munmap() to return a VMA list which is in the original
order - ascending virtual addresses.

Oh, and rename it to <hugh>detach_vmas_to_be_unmapped()</hugh>.  It now
returns nothing, because we know that the VMA we passed in is the head of the
to-be-unmapped list.

22 years ago[PATCH] replace `typedef mmu_gather_t' with `struct mmu_gather'
Andrew Morton [Sat, 11 Jan 2003 02:39:51 +0000 (18:39 -0800)]
[PATCH] replace `typedef mmu_gather_t' with `struct mmu_gather'

In the next patch I wish to add to mm.h prototypes of functions which take an
mmu_gather_t* argument.   To do this I must either:

a) include tlb.h in mm.h

   Not good - more nested includes when a simple forward decl is sufficient.

b) Add `typedef struct free_pte_ctx mmu_gather_t;' to mm.h.

   That's silly - it's supposed to be an opaque type.

   or

c) Remove the pesky typedef.

   Bingo.

22 years ago[PATCH] simplify and generalise cond_resched_lock
Andrew Morton [Sat, 11 Jan 2003 02:39:43 +0000 (18:39 -0800)]
[PATCH] simplify and generalise cond_resched_lock

cond_resched_lock() _used_ to be "if this is the only lock which I am holding
then drop it and schedule if needed".

However with the i_shared_lock->i_shared_sem change, neither of its two
callsites now need those semantics.  So this patch changes it to mean just
"if needed, drop this lock and reschedule".

This allows us to also schedule if CONFIG_PREEMPT=n, which is useful -
zap_page_range() can run for an awfully long time.

The preempt and non-preempt versions of cond_resched_lock() have been
unified.

22 years ago[PATCH] turn i_shared_lock into a semaphore
Andrew Morton [Sat, 11 Jan 2003 02:39:36 +0000 (18:39 -0800)]
[PATCH] turn i_shared_lock into a semaphore

i_shared_lock is held for a very long time during vmtruncate() and causes
high scheduling latencies when truncating a file which is mmapped.  I've seen
100 milliseconds.

So turn it into a semaphore.  It nests inside mmap_sem.

This change is also needed by the shared pagetable patch, which needs to
unshare pte's on the vmtruncate path - lots of pagetable pages need to
be allocated and they are using __GFP_WAIT.

The patch also makes unmap_vma() static.

22 years agoPPC32: Explicitly control store-gathering on MPC10x host bridges.
Randy Vinson [Sat, 11 Jan 2003 02:29:29 +0000 (13:29 +1100)]
PPC32: Explicitly control store-gathering on MPC10x host bridges.

This change alters the behavior of the "MCP10x Store Gathering
Enable" option.  Previously, selecting this option would enable
store gathering, while not selecting the option would leave the
firmware-provided state. In short, there was no way to disable
store gathering if the firmware had left it enabled. This caused
problems with the Adaptec 2940U2W when run in 8240 and 8245
Sandpoint systems. The new code will disable store gathering unless
the user requests it.

22 years ago[PATCH] ptrace-fix-2.5.56-A0
Ingo Molnar [Sat, 11 Jan 2003 02:16:32 +0000 (18:16 -0800)]
[PATCH] ptrace-fix-2.5.56-A0

This patch from Roland McGrath fixes a threading related ptrace bug:
PTRACE_ATTACH should not stop everybody for each thread attached.