]> git.hungrycats.org Git - linux/log
linux
23 years agoUpdate version v2.5.7
Linus Torvalds [Mon, 18 Mar 2002 03:40:24 +0000 (19:40 -0800)]
Update version

23 years agoInclude <linux/completion.h> for completion user
Linus Torvalds [Mon, 18 Mar 2002 03:32:44 +0000 (19:32 -0800)]
Include <linux/completion.h> for completion user

23 years ago[PATCH] cleanup FAT stuff (2/2)
Hirofumi Ogawa [Mon, 18 Mar 2002 03:00:25 +0000 (19:00 -0800)]
[PATCH] cleanup FAT stuff (2/2)

This patch remove unused variable/function/define, and small indent
cleanup.

23 years ago[PATCH] Fix linux/msdos_fs.h for userland (1/2)
Hirofumi Ogawa [Mon, 18 Mar 2002 03:00:01 +0000 (19:00 -0800)]
[PATCH] Fix linux/msdos_fs.h for userland (1/2)

The following patch moves MSDOS_SB() and MSDOS_I() into #define
__KERNEL__.

23 years ago[PATCH] 2.5.7-pre2 IDE 22a
Martin Dalecki [Mon, 18 Mar 2002 02:57:21 +0000 (18:57 -0800)]
[PATCH] 2.5.7-pre2 IDE 22a

- Apply more patches from Vojtech Pavlik for the handling of host chip setup.
   Hopefully they are settled now.

- Kill unused CONFIG_BLK_DEV_MODES

- Push register addressing down in to task_vlb_sync.

- Make the taskfile parsing stuff actually readable. This is compressing the
   code by an incredible amount. We use just one function doing the whole
   scanning right now. This should make sure that the IRQ handler used by a
   particular command is always right.  I didn't introduce typos hopefully
   here.

- Don't call ide_handler_parser as argument for do_taskfile() any longer. We
   have killed this function by coalescing it's functionality with
   ide_cmd_type_parser() anyway.

- Kill unused SLC90E66 code, which Vojtech apparently missed in his patch.

- sync up with 2.5.7-pre2

Once again the actual patch is rather big mostly due to the removal of
some default configuration variables which are not used anylonger. So time for
the next patch stage.

23 years ago[PATCH] for 2.5.7pre2
Jaroslav Kysela [Mon, 18 Mar 2002 02:55:32 +0000 (18:55 -0800)]
[PATCH] for 2.5.7pre2

- add joystick support for CS46xx driver
- Audigy code updates
- fix sound/core/Config.in (wrong dep_tristate usage)
- rawmidi interface fixes (memory leak)
- chang spinlock to rwlock in pcm_native.c (streams linking)
- further fixes of dependencies in Makefiles
- remove experimental time-sync support from sequencer
- fix/update for 32-bit -> 64-bit ioctl converter code
- wavefront driver cleanups
- CMIPCI driver updates
- update joystick support in CS4281
- add detection (not support) of M Audio Delta1010LT
- add AMD768 PCI ID to intel8x0 driver
- add joystick code to trident driver
- remove static variable initialization to zero

23 years ago[PATCH] driverfs support for ISAPNP driver
Andrey Panin [Mon, 18 Mar 2002 02:52:08 +0000 (18:52 -0800)]
[PATCH] driverfs support for ISAPNP driver

This adds initial driverfs support to ISAPNP driver.  It was approved by
the ISAPNP maintainer (Jaroslav Kysela).

23 years ago[PATCH] struct super_block cleanup - affs
Brian Gerst [Mon, 18 Mar 2002 02:50:03 +0000 (18:50 -0800)]
[PATCH] struct super_block cleanup - affs

Seperates affs_sb_info from struct super_block.

23 years ago[PATCH] struct super_block cleanup - hfs
Brian Gerst [Mon, 18 Mar 2002 02:49:31 +0000 (18:49 -0800)]
[PATCH] struct super_block cleanup - hfs

Seperates hfs_sb_info from struct super_block.

23 years ago[PATCH] struct super_block cleanup - shmem
Brian Gerst [Mon, 18 Mar 2002 02:48:42 +0000 (18:48 -0800)]
[PATCH] struct super_block cleanup - shmem

Seperates shmem_sb_info from struct super_block.

23 years agoAutomerge
Linus Torvalds [Mon, 18 Mar 2002 02:47:47 +0000 (18:47 -0800)]
Automerge

23 years ago[PATCH] moving stuff to fs/filesystems.c
Alexander Viro [Mon, 18 Mar 2002 02:42:06 +0000 (18:42 -0800)]
[PATCH] moving stuff to fs/filesystems.c

file_system_typer-related code moved from fs/super.c to
fs/filesystems.c

23 years agoMerge http://linuxusb.bkbits.net/linus-2.5
Linus Torvalds [Mon, 18 Mar 2002 02:41:50 +0000 (18:41 -0800)]
Merge http://linuxusb.bkbits.net/linus-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

23 years agoAutomerge
Linus Torvalds [Mon, 18 Mar 2002 02:39:06 +0000 (18:39 -0800)]
Automerge

23 years ago[PATCH] struct super_block cleanup - udf
Brian Gerst [Mon, 18 Mar 2002 02:38:40 +0000 (18:38 -0800)]
[PATCH] struct super_block cleanup - udf

Seperates udf_sb_info from struct super_block.

23 years ago[PATCH] struct super_block cleanup - isofs
Brian Gerst [Mon, 18 Mar 2002 02:38:35 +0000 (18:38 -0800)]
[PATCH] struct super_block cleanup - isofs

Seperates isofs_sb_info from struct super_block.

23 years ago[PATCH] PATCH -- pci_pool and CONFIG_DEBUG_SLAB
David Brownell [Mon, 18 Mar 2002 02:37:33 +0000 (18:37 -0800)]
[PATCH] PATCH -- pci_pool and CONFIG_DEBUG_SLAB

I got burnt one too many time by mismatches between
the pci_pool and "real" slabs... something changed in
mm/slab.c and broke a driver, so I'm going for the real
fix this time.  Having poisoning that _works_ is a huge
help in the innards of the USB host controller drivers.

This patch gets rid of some #ifdefs and makes the pci_pool
code poison memory if CONFIG_DEBUG_SLAB is set.
The functionality has always been there, but this makes
it simpler to get at.

23 years agoMerge bk://ppc.bkbits.net/for-linus-ppc
Linus Torvalds [Mon, 18 Mar 2002 02:15:57 +0000 (18:15 -0800)]
Merge bk://ppc.bkbits.net/for-linus-ppc
into home.transmeta.com:/home/torvalds/v2.5/linux

23 years agoFix a lockup on some PPC machines running an SMP kernel - we were
Paul Mackerras [Tue, 19 Mar 2002 09:40:08 +0000 (20:40 +1100)]
Fix a lockup on some PPC machines running an SMP kernel - we were
exiting heathrow_modem_enable() with a lock held.

23 years agoPPC update - add preempt_count to the ppc thread_info, add
Paul Mackerras [Tue, 19 Mar 2002 09:37:54 +0000 (20:37 +1100)]
PPC update - add preempt_count to the ppc thread_info, add
SI_DETHREAD, plus a couple of minor fixes.

23 years agoMerge master.kernel.org:/home/davem/BK/net-2.5
Linus Torvalds [Mon, 18 Mar 2002 02:06:55 +0000 (18:06 -0800)]
Merge master.kernel.org:/home/davem/BK/net-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

23 years agoSwitch cramfs and zisofs from zlib_fs to common zlib.
David Woodhouse [Sun, 17 Mar 2002 20:02:32 +0000 (20:02 +0000)]
Switch cramfs and zisofs from zlib_fs to common zlib.
Remove remnants of zlib_fs.

23 years agoRemove jffs2_sb from struct super_block union.
David Woodhouse [Sun, 17 Mar 2002 15:38:00 +0000 (15:38 +0000)]
Remove jffs2_sb from struct super_block union.
Remove FS_REQUIRES_DEV from JFFS2. We never really used the block device anyway.

23 years agoMake wanpipe build again after struct sock cleanups.
David S. Miller [Sun, 17 Mar 2002 12:03:52 +0000 (04:03 -0800)]
Make wanpipe build again after struct sock cleanups.
From Arnaldo Carvalho de Melo.

23 years agoFix netfilter IPv4 conntrack build.
David S. Miller [Sun, 17 Mar 2002 11:55:42 +0000 (03:55 -0800)]
Fix netfilter IPv4 conntrack build.

23 years agoUSB printer patch
Taisuke Yamada [Sun, 17 Mar 2002 02:48:38 +0000 (18:48 -0800)]
USB printer patch

added NEC printer to quirks list

23 years agoUSB HID driver
Vojtech Pavlik [Sun, 17 Mar 2002 02:38:43 +0000 (18:38 -0800)]
USB HID driver

Workaround for the ATEN switches

23 years agoUSB serial drivers
Ganesh Varadarajan [Sun, 17 Mar 2002 02:31:22 +0000 (18:31 -0800)]
USB serial drivers

Several functions in the serial drivers can be called from bottom
half or interrupt context. They must use the GFP_ATOMIC flag for
calls to kmalloc() and usb_submit_urb().

Functions which must use GFP_ATOMIC:
1. All *_callback() functions.
2. Any code which is inside a spinlock.
3. write(), throttle(), unthrottle(), which may be called by
   the line discipline in bottom half context.

Functions which can use GFP_KERNEL:
1. open(), close(), startup(), shutdown(), set_termios().

23 years agoUSB catc driver
Simon Evans [Sun, 17 Mar 2002 02:29:11 +0000 (18:29 -0800)]
USB catc driver

Here is a patch to add support for F5U011 to catc.c driver. The
patch has been compile tested against 2.5.6 and 2.5.7pre1
(and tested against 2.5.5-dj1) and should apply cleanly.

23 years agoUSB Urefs for hid-core/hiddev
Paul Stewart [Sun, 17 Mar 2002 02:14:24 +0000 (18:14 -0800)]
USB Urefs for hid-core/hiddev

  I've written a patch Vojtech and I discussed for enhancing the
hiddev code to optionally provide more detailed output on read().
The old functionality is still supported by default, and in
situations where HID usage codes are unique across reports, the
old method is still preferable due to its terseness.

  The new method provides the ability to determine exactly which
value has changed, in cases where the HID usage codes are not
unique.  It also provides a means to optionally receive notification
when input reports are received from the device, whether or not
any of the values in the report have changed.

  The details of the changes are as follows:

  - All current code behaves identically

  - A new ioctl pair HIDIOCGFLAG/HIDIOCSFLAG gets and clears
    flags on the hiddev device.

  - If you set the flag HIDDEV_FLAG_UREF, the read() call switches
    from reading hiddev_event structures to hiddev_usage_ref
    structures.  The change takes effect immediately, even to
    already queued events that haven't been read() yet.  Here's
    an example of enabling FLAG_UREF:

    {
        int flag = HIDDEV_FLAG_UREF;
        if (ioctl(fd, HIDIOCSFLAG, &flag) != 0) {
                perror("ioctl");
                exit(1);
        }
    }

  - With the HIDDEV_FLAG_REPORT set (which is only allowed if
    HIDDEV_FLAG_UREF is also set), there is a special uref that
    will be read() in addition to the ones corresponding to
    changes in the device state: when uref.field_index is set to
    HID_FIELD_INDEX_NONE, this uref is a notification that the
    report referred to by report_type and report_id has been
    received from the device.  This can be useful in situations
    when the notification of the arrival of a report is useful
    even if there is no change in state.

23 years agoUSB printer update
David Paschal [Sun, 17 Mar 2002 02:07:21 +0000 (18:07 -0800)]
USB printer update

- bind to 7/1/2 alternate setting by default, to fix printing with HP
  LaserJet 1200 and 2200
- ioctls needed by the GPL user-mode IEEE 1284.4 driver which is part of
  the HP OfficeJet Linux driver (http://hpoj.sourceforge.net):
  - dynamic switching between 7/1/[123] alternate settings
  - sending HP vendor-specific channel-change-request to support
    memory card readers on HP PhotoSmart printers
  - inquire more information about the peripheral, including
    /proc/bus/usb/xx/yy linkage to get even more information
- fix apparent array overflow (by 1 entry) in usblp_probe when more than
  the maximum number of USB printers are connected
- for the 2.2 version, added MODULE_{INC,DEC}_USE_COUNT to prevent rmmoding
  of printer.o (and subsequent OOPSes) while a USB printer device is open
- cleaned up the code in a few places by consolidating duplicated code

23 years agoMerge http://linux-isdn.bkbits.net/linux-2.5.misc
Linus Torvalds [Sat, 16 Mar 2002 08:03:06 +0000 (00:03 -0800)]
Merge http://linux-isdn.bkbits.net/linux-2.5.misc
into home.transmeta.com:/home/torvalds/v2.5/linux

23 years agoMerge http://linux-isdn.bkbits.net/linux-2.5.make
Linus Torvalds [Sat, 16 Mar 2002 08:02:34 +0000 (00:02 -0800)]
Merge http://linux-isdn.bkbits.net/linux-2.5.make
into home.transmeta.com:/home/torvalds/v2.5/linux

23 years agoDescend into drivers/hotplug only if CONFIG_HOTPLUG_PCI is set.
Kai Germaschewski [Sat, 16 Mar 2002 11:47:49 +0000 (05:47 -0600)]
Descend into drivers/hotplug only if CONFIG_HOTPLUG_PCI is set.

23 years agoDescend into drivers/parport only if CONFIG_PARPORT is set.
Kai Germaschewski [Sat, 16 Mar 2002 11:45:45 +0000 (05:45 -0600)]
Descend into drivers/parport only if CONFIG_PARPORT is set.

23 years agoLink drivers/fc4/fc4.a only once.
Kai Germaschewski [Sat, 16 Mar 2002 11:43:26 +0000 (05:43 -0600)]
Link drivers/fc4/fc4.a only once.

23 years agoFix up ACPI so that it seems to work in the new world order:
Linus Torvalds [Sat, 16 Mar 2002 07:58:39 +0000 (23:58 -0800)]
Fix up ACPI so that it seems to work in the new world order:
make driverfs initialize early, so that ACPI can come alive
in a world where you can register devices.

23 years agoMerge tp1.ruhr-uni-bochum.de:/home/kai/kernel/v2.5/linux-2.5.config_sound
Kai Germaschewski [Fri, 15 Mar 2002 17:04:34 +0000 (11:04 -0600)]
Merge tp1.ruhr-uni-bochum.de:/home/kai/kernel/v2.5/linux-2.5.config_sound
into tp1.ruhr-uni-bochum.de:/home/kai/kernel/v2.5/linux-2.5.config_sound.x

23 years agoupdate v2.5.7-pre2
Linus Torvalds [Fri, 15 Mar 2002 09:46:28 +0000 (01:46 -0800)]
update

23 years agoUpdate kernel version
Linus Torvalds [Fri, 15 Mar 2002 09:44:59 +0000 (01:44 -0800)]
Update kernel version

23 years agoMake sk_flags unsigned long, since we do bit operations on it
Linus Torvalds [Fri, 15 Mar 2002 09:44:42 +0000 (01:44 -0800)]
Make sk_flags unsigned long, since we do bit operations on it

23 years agoFix up ACPI device breakage.
Linus Torvalds [Fri, 15 Mar 2002 09:43:33 +0000 (01:43 -0800)]
Fix up ACPI device breakage.

For some reason the ACPI people continue to make the mistake
of thinking that they are the root of the system.  Disabuse
them of that notion.

23 years ago[PATCH] 2.5.7-pre1 Code cleanup for BSD accounting.
Bob Miller [Fri, 15 Mar 2002 08:15:08 +0000 (00:15 -0800)]
[PATCH] 2.5.7-pre1 Code cleanup for BSD accounting.

Clean up BSD accounting locking code..

23 years agoTrivial compile fix
Linus Torvalds [Fri, 15 Mar 2002 08:06:42 +0000 (00:06 -0800)]
Trivial compile fix

23 years ago[PATCH] sync shmem.c in 2.5 to 2.4
Christoph Rohland [Fri, 15 Mar 2002 08:00:39 +0000 (00:00 -0800)]
[PATCH] sync shmem.c in 2.5 to 2.4

The appended patch brings the fixes applied in 2.4 to shmem.c to 2.5.

In Detail:
- Add needed checks for shmem_file_write and shmem_symlink
- Add Documentation/filesystems/tmpfs.txt and adjust Config.help
- Add uid and gid mount options
- Make the error messages more user friendly

23 years ago[PATCH] binfmt_elf.c: do SET_PERSONALITY() for static binaries
David Mosberger [Fri, 15 Mar 2002 07:45:32 +0000 (23:45 -0800)]
[PATCH] binfmt_elf.c: do SET_PERSONALITY() for static binaries

Pick up binfmt_elf.c SET_PERSONALITY() fix from 2.4.18.

23 years agoFix overenthusiastic ia64 merge.
Linus Torvalds [Fri, 15 Mar 2002 07:43:42 +0000 (23:43 -0800)]
Fix overenthusiastic ia64 merge.

That preempt_count really is supposed to be unconditional,
architectures please take note and add to your thread info.

23 years agoMerge bk://lia64.bkbits.net/to-linus-2.5/
Linus Torvalds [Fri, 15 Mar 2002 07:35:12 +0000 (23:35 -0800)]
Merge bk://lia64.bkbits.net/to-linus-2.5/
into home.transmeta.com:/home/torvalds/v2.5/linux

23 years ago[PATCH] ACPI patch 9/9
Andy Grover [Fri, 15 Mar 2002 07:28:59 +0000 (23:28 -0800)]
[PATCH] ACPI patch 9/9

If you could only review one of the 9 patches, this would be the one.

- removes acpitable.c vestiges
- adds ACPI IRQ routing support to PCI (disableable via pci=noacpi option)
- adds code to get a <1MB page for sleep, and ACPI boot to setup.c
- allocates another page in the fixmap for ACPI
- changes driverfs a little to work better with ACPI.

23 years ago[PATCH] ACPI patch 8/9
Andy Grover [Fri, 15 Mar 2002 07:28:54 +0000 (23:28 -0800)]
[PATCH] ACPI patch 8/9

This patch removes arch/i386/kernel/acpitable.c. As mentioned previously,
the new ACPI code integrates this, so it's no longer needed.

23 years ago[PATCH] ACPI patch 7/9
Andy Grover [Fri, 15 Mar 2002 07:28:50 +0000 (23:28 -0800)]
[PATCH] ACPI patch 7/9

This updates the Configure.help, both in arch/i386, and in drivers/acpi.

23 years ago[PATCH] ACPI patch 6/9
Andy Grover [Fri, 15 Mar 2002 07:28:46 +0000 (23:28 -0800)]
[PATCH] ACPI patch 6/9

This removes the old OSPM code. It lived under drivers/acpi/ospm/*, but
the new code just lives in drivers/acpi, and removes some unnecessary
abstraction that this old code had.

23 years ago[PATCH] ACPI patch 5/9
Andy Grover [Fri, 15 Mar 2002 07:28:40 +0000 (23:28 -0800)]
[PATCH] ACPI patch 5/9

This is the update to the core interpreter code.

23 years ago[PATCH] ACPI patch 4/9
Andy Grover [Fri, 15 Mar 2002 07:28:32 +0000 (23:28 -0800)]
[PATCH] ACPI patch 4/9

This is the config.in and makefile changes for the latest code. The most
(only) interesting thing probably is ACPI is no longer flagged experimental.

23 years ago[PATCH] ACPI patch 3/9
Andy Grover [Fri, 15 Mar 2002 07:28:27 +0000 (23:28 -0800)]
[PATCH] ACPI patch 3/9

This patch updates the acpi IA32 arch-specific files. Part of this is
taking what was acpitable.c and implementing it with better integration
with the rest of the ACPI code.

23 years ago[PATCH] ACPI patch 2/9
Andy Grover [Fri, 15 Mar 2002 07:28:23 +0000 (23:28 -0800)]
[PATCH] ACPI patch 2/9

This patch adds in the new drivers.

- Support for driverfs
- File/code layout more in the Linux style
- improvements to battery, processor, and thermal support

23 years ago[PATCH] ACPI patch 1/9
Andy Grover [Fri, 15 Mar 2002 07:28:19 +0000 (23:28 -0800)]
[PATCH] ACPI patch 1/9

This is the first of 9 patches. We did a complete rewrite of the
Linux-specific code, so we wait for things to stabilize before submitting.
There will be more updates, but *much* smaller.

#1 - this updates the header file.

23 years ago[PATCH] boot_cpu_data corruption on SMP x86
Mikael Pettersson [Fri, 15 Mar 2002 07:22:10 +0000 (23:22 -0800)]
[PATCH] boot_cpu_data corruption on SMP x86

The patch below eliminates a case of boot_cpu_data corruption
on SMP x86 machines. This was first observed on SMP Athlons,
but it also affects SMP Intel boxes in a less serious way.

When the secondary processors boot and execute head.S:checkCPUtype,
the code performs a 32-bit write of a small constant to the
byte-sized variable boot_cpu_data.x86 (X86 in head.S). Since the
write is 32-bit, it also writes zeros to the following 3 bytes,
which clobbers the x86_vendor, x86_model, and x86_mask fields
previously set up by check_bugs()'s call to identify_cpu().
Thus, after smp_init(), boot_cpu_data will _always_ identify
the CPU as an Intel (X86_VENDOR_INTEL == 0 in processor.h) with
model 0 and stepping 0.

The effect in standard kernels is not catastrophic, since:
(a) most SMP x86 boxes are Intel
(b) most uses of x86_vendor occur before smp_init() or reference
    the SMP cpu_data[] array
(c) most post-boot references to boot_cpu_data occur in the
    cpu_has_XXX macros which only read the x86_capability[] array
However, third-party extensions (like my x86 performance-monitoring
conters driver) can get seriously confused by this mis-identification.

23 years agoAutomerge
Linus Torvalds [Fri, 15 Mar 2002 07:19:54 +0000 (23:19 -0800)]
Automerge

23 years ago[PATCH] struct super_block cleanup - msdos/vfat
Brian Gerst [Fri, 15 Mar 2002 07:16:12 +0000 (23:16 -0800)]
[PATCH] struct super_block cleanup - msdos/vfat

Don't print out FAT superblock warnings if the IO failed.

23 years ago[PATCH] struct super_block cleanup - qnx4
Brian Gerst [Fri, 15 Mar 2002 07:16:08 +0000 (23:16 -0800)]
[PATCH] struct super_block cleanup - qnx4

Seperates qnx4_sb_info from struct super_block.

23 years ago[PATCH] struct super_block cleanup - smbfs
Brian Gerst [Fri, 15 Mar 2002 07:16:03 +0000 (23:16 -0800)]
[PATCH] struct super_block cleanup - smbfs

Seperates smb_sb_info from struct super_block.

23 years ago[PATCH] struct super_block cleanup - msdos/vfat
Brian Gerst [Fri, 15 Mar 2002 07:15:59 +0000 (23:15 -0800)]
[PATCH] struct super_block cleanup - msdos/vfat

Seperates msdos_sb_info from struct super_block for msdos and vfat.
Umsdos is terminally broken and is not included.

23 years ago[PATCH] Fix NR_IRQS when no IO apic
Brian Gerst [Fri, 15 Mar 2002 07:15:55 +0000 (23:15 -0800)]
[PATCH] Fix NR_IRQS when no IO apic

NR_IRQS should be 16 when the IO apic is not configured, as the 8259 PIC
cannot generate any more interrupts.  It also fixes a bug where the IDT
gets populated with random addresses, since only 16 entry stubs are
created.

23 years ago[PATCH] Cleanup F00F bug code
Brian Gerst [Fri, 15 Mar 2002 07:15:51 +0000 (23:15 -0800)]
[PATCH] Cleanup F00F bug code

This changes the F00F bug workaround code to use the fixmap facilities
instead of touching the page tables directly.  It also removes the
assumption that only 686's don't have the bug.  I have confirmation that
the patch works on buggy pentiums.

23 years agoFix up architectures for task list changes
Linus Torvalds [Fri, 15 Mar 2002 07:12:06 +0000 (23:12 -0800)]
Fix up architectures for task list changes

23 years ago[PATCH] nfsd-as-fs NULL ptr fix
Alexander Viro [Fri, 15 Mar 2002 07:06:14 +0000 (23:06 -0800)]
[PATCH] nfsd-as-fs NULL ptr fix

It needs the following patch

23 years ago[PATCH] proc_pid_make_inode() fix
Alexander Viro [Fri, 15 Mar 2002 07:06:10 +0000 (23:06 -0800)]
[PATCH] proc_pid_make_inode() fix

In case if proc_pid_make_inode() steps on exiting task we do
iput() and return NULL.  Unfortunately, in that case inode->i_ino
doesn't look like inumber of a per-process inode and we take the
wrong path in proc_delete_inode().  I.e. do dput(PDE(inode)).  Which
is left uninitialized...

We used to get out with that almost by accident - that code
worked only because we had zeroed out one field of union and that
guaranteed that another field would be NULL.  It worked, but broke
at the first occasion.

23 years ago[PATCH] nfsd as filesystem
Alexander Viro [Fri, 15 Mar 2002 07:06:06 +0000 (23:06 -0800)]
[PATCH] nfsd as filesystem

* introduces a new filesystem - nfsd.  No, it's not a typo.  It's a small
  tree with fixed topology defined by nfsd and IO on its files does what
  we used to do by hand in nfsctl.c.
* turns sys_nfsservctl() into a sequence of open()/write()/read()/close()
  It works as it used to - we don't need nfsd to be mounted anywhere, etc.
* nfsd_linkage ugliness is gone.
* getfs and getfh demonstrate (rather trivial) example of "descriptor as
  transaction descriptor" behaviour.
* we are fairly close to the situation when driver-defined filesystems can
  be done with practically zero code overhead.  We are still not there, but
  it's a matter of adding a couple of helpers for populating the tree.

One thing we get immediately is a cleanup of sys_nfsservctl() -
it got _much_ better.  Moreover, we get an alternative interface that
uses normal file IO and can be used without magic syscalls.

23 years ago[PATCH] fix for leaks in nfsd
Alexander Viro [Fri, 15 Mar 2002 07:06:02 +0000 (23:06 -0800)]
[PATCH] fix for leaks in nfsd

Several exits in exp_export() forget to call path_release().  Fixed.

23 years ago[PATCH] fs/libfs.c
Alexander Viro [Fri, 15 Mar 2002 07:05:58 +0000 (23:05 -0800)]
[PATCH] fs/libfs.c

Linus, I've taken a bunch of common methods into fs/libfs.c and
killed the (duplicated) instances in filesystems.  There will be more -
ideally I'd like to get a library that would make writing small filesystems
trivial.

23 years agoMerge master.kernel.org:/home/davem/BK/net-2.5
Linus Torvalds [Fri, 15 Mar 2002 07:01:09 +0000 (23:01 -0800)]
Merge master.kernel.org:/home/davem/BK/net-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

23 years ago[PATCH] 2.4 and 2.5: fix /proc/kcore
Russell King [Fri, 15 Mar 2002 06:44:26 +0000 (22:44 -0800)]
[PATCH] 2.4 and 2.5: fix /proc/kcore

As mentioned on May 11 on LKML, here is a patch to fix /proc/kcore for
architectures which do not have RAM located at physical address 0.

23 years ago[PATCH] 2.4 and 2.5: remove Alt-Sysrq-L
Russell King [Fri, 15 Mar 2002 06:44:22 +0000 (22:44 -0800)]
[PATCH] 2.4 and 2.5: remove Alt-Sysrq-L

The following patch removes Alt-Sysrq-L and its associated hack to kill
of PID1, the init process.  This is a mis-feature.

If PID1 is killed, the kernel immediately enters an infinite loop in the
depths of do_exit() with interrupts disabled, completely locking the
machine.  Obviously you can only reach for the reset button or power
switch after this, leaving you with dirty filesystems.

23 years ago[PATCH] 2.5 and 2.4: fix PCI IO BAR flags
Russell King [Fri, 15 Mar 2002 06:44:18 +0000 (22:44 -0800)]
[PATCH] 2.5 and 2.4: fix PCI IO BAR flags

There is a problem where the resource flags sometimes contain bits from
the address part of the PCI BAR, especially when you have the low address
bit set for an IO BAR.

(bit 3 of a PCI IO BAR is an address bit, and (bar & 0xf) propagates this
to res->flags).

This exists in Ivan Kokshaysky PCI patches, and so far hasn't made it into
the kernel.  It's required for IDE on certain ARM machines to even work.

23 years ago[PATCH] wait4() WIFSTOPPED starvation fix #2/2
David Howells [Fri, 15 Mar 2002 06:39:17 +0000 (22:39 -0800)]
[PATCH] wait4() WIFSTOPPED starvation fix #2/2

This patch actually fixes the starvation bug in sys_wait4() by moving any
process which is serviced for stoppage to the end of the child list.

23 years agoCleanup: use list macros for task list
Linus Torvalds [Fri, 15 Mar 2002 06:30:56 +0000 (22:30 -0800)]
Cleanup: use list macros for task list

23 years ago[PATCH] wait4() WIFSTOPPED starvation fix #1/2
David Howells [Fri, 15 Mar 2002 05:20:04 +0000 (21:20 -0800)]
[PATCH] wait4() WIFSTOPPED starvation fix #1/2

This patch (#1) just converts the task_struct to use struct list_head rather
than direct pointers for maintaining the children list.

23 years agoMerge linux-isdn@linux-isdn.bkbits.net:linux-2.5.make
Kai Germaschewski [Fri, 15 Mar 2002 08:26:53 +0000 (02:26 -0600)]
Merge linux-isdn@linux-isdn.bkbits.net:linux-2.5.make
into tp1.ruhr-uni-bochum.de:/home/kai/kernel/v2.5/linux-2.5.make

23 years agoRemove link rules for multi-part drivers in drivers/isdn/*/Makefile.
Kai Germaschewski [Fri, 15 Mar 2002 08:25:37 +0000 (02:25 -0600)]
Remove link rules for multi-part drivers in drivers/isdn/*/Makefile.

23 years agoImprove Rules.make to automatically generate link rules for composite
Kai Germaschewski [Fri, 15 Mar 2002 08:19:57 +0000 (02:19 -0600)]
Improve Rules.make to automatically generate link rules for composite
objects.

Current behavior is not changed at all, but see the next cset for what
it's good for.

23 years agoFix drivers/pnp/Makefile to correctly list multi-part
Kai Germaschewski [Fri, 15 Mar 2002 08:09:52 +0000 (02:09 -0600)]
Fix drivers/pnp/Makefile to correctly list multi-part
objects in $(list-multi), instead of $(multi-objs)

23 years ago[PATCH] Fix IA-32 Intercept code.
Junichi Nomura [Fri, 15 Mar 2002 04:06:28 +0000 (20:06 -0800)]
[PATCH] Fix IA-32 Intercept code.

(ia32_intercept): The code for Locked data reference fault is 4, not 3.

23 years agoMerge http://gkernel.bkbits.net/misc-2.5
Linus Torvalds [Fri, 15 Mar 2002 01:23:30 +0000 (17:23 -0800)]
Merge http://gkernel.bkbits.net/misc-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

23 years agoAdd BK kernel howto text and some helper scripts to
Jeff Garzik [Fri, 15 Mar 2002 07:10:28 +0000 (02:10 -0500)]
Add BK kernel howto text and some helper scripts to
new subdirectory linux/Documentation/BK-usage.

23 years agoMerge http://gkernel.bkbits.net/misc-2.5
Linus Torvalds [Fri, 15 Mar 2002 00:54:30 +0000 (16:54 -0800)]
Merge http://gkernel.bkbits.net/misc-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

23 years agoMerge http://gkernel.bkbits.net/net-drivers-2.5
Linus Torvalds [Fri, 15 Mar 2002 00:54:00 +0000 (16:54 -0800)]
Merge http://gkernel.bkbits.net/net-drivers-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

23 years agoFix e1000 net driver build with newer binutils.
Jeff Garzik [Thu, 14 Mar 2002 23:44:17 +0000 (18:44 -0500)]
Fix e1000 net driver build with newer binutils.

23 years agoDon't include linux/delay.h twice in eepro100 net driver.
Jeff Garzik [Thu, 14 Mar 2002 23:42:48 +0000 (18:42 -0500)]
Don't include linux/delay.h twice in eepro100 net driver.

Noticed by Alan Cox.

23 years agoConvert hp100 net driver to PCI DMA mapping API. (fixes build)
Jean Tourrilhes [Thu, 14 Mar 2002 23:07:15 +0000 (18:07 -0500)]
Convert hp100 net driver to PCI DMA mapping API.  (fixes build)

23 years agoAdd new slicecom/munish WAN driver.
Szilárd Pásztor [Thu, 14 Mar 2002 23:00:03 +0000 (18:00 -0500)]
Add new slicecom/munish WAN driver.

23 years agoFix via audio recording, when frag size < page size.
Itai Nahshon [Thu, 14 Mar 2002 22:55:38 +0000 (17:55 -0500)]
Fix via audio recording, when frag size < page size.

23 years agolance net driver update: mark lance_probe as __init
Paul Gortmaker [Thu, 14 Mar 2002 22:51:52 +0000 (17:51 -0500)]
lance net driver update:  mark lance_probe as __init

23 years agoJanitor: request_region cleanups for stallion serial driver
William Stinson [Thu, 14 Mar 2002 22:50:10 +0000 (17:50 -0500)]
Janitor: request_region cleanups for stallion serial driver

23 years agoFix rocketport serial driver for kdev_t changes in early 2.5.x series.
William Stinson [Thu, 14 Mar 2002 22:48:36 +0000 (17:48 -0500)]
Fix rocketport serial driver for kdev_t changes in early 2.5.x series.

23 years agoclgenfb fixes for zorro bus. clgenfb should work again on m68k.
linux-m68k.org@mandrakesoft.com [Thu, 14 Mar 2002 22:46:56 +0000 (17:46 -0500)]
clgenfb fixes for zorro bus.  clgenfb should work again on m68k.

23 years agoacenic gige net driver update: merge VLAN support from 2.4.x kernel
Jeff Garzik [Thu, 14 Mar 2002 22:35:21 +0000 (17:35 -0500)]
acenic gige net driver update: merge VLAN support from 2.4.x kernel

23 years agoacenic gige net driver fixes:
Jes Sorensen [Thu, 14 Mar 2002 22:32:21 +0000 (17:32 -0500)]
acenic gige net driver fixes:
* fix Tigon I support
* fix memory leak

23 years agoacenic gige net driver updates:
Jes Sorensen [Thu, 14 Mar 2002 22:19:12 +0000 (17:19 -0500)]
acenic gige net driver updates:
* various small cleanups
* ETHTOOL_GDRVINFO support

23 years agoMerge mandrakesoft.com:/home/jgarzik/repo/anton-pcnet32-2.5
Jeff Garzik [Thu, 14 Mar 2002 21:54:59 +0000 (16:54 -0500)]
Merge mandrakesoft.com:/home/jgarzik/repo/anton-pcnet32-2.5
into mandrakesoft.com:/home/jgarzik/repo/net-drivers-2.5