openwrt/staging/blogic.git
19 years ago[ARM] 2876/1: N30 - remove task to toggle USB D+ line
Ben Dooks [Mon, 5 Sep 2005 19:47:53 +0000 (20:47 +0100)]
[ARM] 2876/1: N30 - remove task to toggle USB D+ line

Patch from Ben Dooks

The n30 machine file is using a kernel thread to change the
state of the USB D+ pull-up resistor, which is not the proper
way to do this. Once the usb-gadget support for the 24xx is
merge, the proper access method will be added.
This patch also removes the problem of using HZ for the
msleep calls, from Nishanth Aravamudan.

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] Remove unused DYN_TICK_* macros
Russell King [Mon, 5 Sep 2005 09:21:04 +0000 (10:21 +0100)]
[ARM] Remove unused DYN_TICK_* macros

Neither DYN_TICK_SKIPPING nor DYN_TICK_SUITABLE are used on ARM.
Remove them.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] Wrap calls to descriptor handlers
Russell King [Sun, 4 Sep 2005 18:45:00 +0000 (19:45 +0100)]
[ARM] Wrap calls to descriptor handlers

This is part of Thomas Gleixner's generic IRQ patch, which converts
ARM to use the generic IRQ subsystem.  Here, we wrap calls to
desc->handler() in an inline function, desc_handle_irq().  This
reduces the size of Thomas' patch since the changes become more
localised.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] Change irq_chip wake/type methods to set_wake/set_type
Russell King [Sun, 4 Sep 2005 18:43:13 +0000 (19:43 +0100)]
[ARM] Change irq_chip wake/type methods to set_wake/set_type

This is part of Thomas Gleixner's generic IRQ patch, which converts
ARM to use the generic IRQ subsystem.  Here, we rename two of the
irq_chip methods - wake becomes set_wake, and type becomes set_type.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] Fix compilation in locomo.c
Pavel Machek [Sun, 4 Sep 2005 10:33:12 +0000 (11:33 +0100)]
[ARM] Fix compilation in locomo.c

Do not access children in struct device directly, use
device_for_each_child helper instead. It fixes compilation.

Signed-off-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] Stack starts at THREAD_START_SP offset, not THREAD_SIZE-8
Russell King [Sun, 4 Sep 2005 10:03:15 +0000 (11:03 +0100)]
[ARM] Stack starts at THREAD_START_SP offset, not THREAD_SIZE-8

Use the correct constants.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] 2875/1: Data Abort fixes
Timothy Baldwin [Sun, 4 Sep 2005 09:13:48 +0000 (10:13 +0100)]
[ARM] 2875/1: Data Abort fixes

Patch from Timothy Baldwin

All data aborts are treated as read accesses. The existing code updates the wrong bit of r1, also the comments are wrong in that the sense of the L bit is inverted.

Signed-off-by: Timothy E. Baldwin <T.E.Baldwin99@members.leeds.ac.uk>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] 2874/1: S3C2410 - add cpu_init() call after sleep wakeup
Ben Dooks [Sat, 3 Sep 2005 18:39:26 +0000 (19:39 +0100)]
[ARM] 2874/1: S3C2410 - add cpu_init() call after sleep wakeup

Patch from Ben Dooks

The power management sleep code needs to call cpu_init()
to restore the cpu state after the system resumes from
suspend. Also clear off an un-necessary comment.
Thanks to Dimitry Andric for reporting the bug and
for rmk for pointing out the cause.

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] 2873/1: PCMCIA soc: Allow access to filesystems on CF at boot time
Richard Purdie [Sat, 3 Sep 2005 18:39:25 +0000 (19:39 +0100)]
[ARM] 2873/1: PCMCIA soc: Allow access to filesystems on CF at boot time

Patch from Richard Purdie

This change makes the soc pcmcia interfaces available earlier in the
boot process meaning devices like CF microdrives can be used for the
root filesystem.

Signed-off-by: Richard Purdie <rpurdie@rpsys.net>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years agoMerge refs/heads/ieee80211-wifi from master.kernel.org:/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 2 Sep 2005 09:01:35 +0000 (02:01 -0700)]
Merge refs/heads/ieee80211-wifi from /linux/kernel/git/jgarzik/netdev-2.6

19 years ago[wireless hostap] automatically select ieee80211 dependency in Kconfig
Jeff Garzik [Fri, 2 Sep 2005 08:44:25 +0000 (04:44 -0400)]
[wireless hostap] automatically select ieee80211 dependency in Kconfig

19 years ago[PATCH] remove driverfs references from init/do_mounts.c
Rolf Eike Beer [Fri, 2 Sep 2005 07:03:09 +0000 (09:03 +0200)]
[PATCH] remove driverfs references from init/do_mounts.c

This patch is against 2.6.10, but still applies cleanly. It's just
s/driverfs/sysfs/ in this file.

Signed-off-by: Rolf Eike Beer <eike-kernel@sf-tec.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] remove driverfs references from include/linux/cpu.h and net/sunrpc/rpc_pipe.c
Rolf Eike Beer [Fri, 2 Sep 2005 06:59:25 +0000 (08:59 +0200)]
[PATCH] remove driverfs references from include/linux/cpu.h and net/sunrpc/rpc_pipe.c

This patch is against 2.6.10, but still applies cleanly. It's just
s/driverfs/sysfs/ in these two files.

Signed-off-by: Rolf Eike Beer <eike-kernel@sf-tec.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v850: Add show_mem
Miles Bader [Fri, 2 Sep 2005 06:13:31 +0000 (15:13 +0900)]
[PATCH] v850: Add show_mem

Signed-off-by: Miles Bader <miles@gnu.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v850: Update defconfigs
Miles Bader [Fri, 2 Sep 2005 06:13:30 +0000 (15:13 +0900)]
[PATCH] v850: Update defconfigs

Signed-off-by: Miles Bader <miles@gnu.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] v850: Round up length passed to slram driver to a multiple of SLRAM_BLK_SZ
Miles Bader [Fri, 2 Sep 2005 06:13:30 +0000 (15:13 +0900)]
[PATCH] v850: Round up length passed to slram driver to a multiple of SLRAM_BLK_SZ

Signed-off-by: Miles Bader <miles@gnu.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] uclinux: use MAP_PRIVATE when mmaping code regions in flat binary loader
Greg Ungerer [Fri, 2 Sep 2005 00:42:52 +0000 (10:42 +1000)]
[PATCH] uclinux: use MAP_PRIVATE when mmaping code regions in flat binary loader

Use MAP_PRIVATE when calling mmap to get memory for the code region.
The flat loader was using MAP_SHARED, but underlying changes to the
MMUless mmap means this is now wrong.

Signed-off-by: Greg Ungerer <gerg@uclinux.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] m68knommu: update defconfig for m68knommu
Greg Ungerer [Fri, 2 Sep 2005 00:42:52 +0000 (10:42 +1000)]
[PATCH] m68knommu: update defconfig for m68knommu

Updated defconfig for m68knommu arch.
Patch originaly submitted by Jan Dittmer <jdittmer@ppp0.net>

Signed-off-by: Greg Ungerer <gerg@uclinux.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] m68knommu: new board support in linker script
Greg Ungerer [Fri, 2 Sep 2005 00:42:52 +0000 (10:42 +1000)]
[PATCH] m68knommu: new board support in linker script

. add support for the M5235EVB board
. add support for the SOM5282 board
. add support for the MOD5272 board
. fix end of memory define for eLITE board

Signed-off-by: Greg Ungerer <gerg@uclinux.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] m68knommu: need pfn_valid macro
Greg Ungerer [Fri, 2 Sep 2005 00:42:52 +0000 (10:42 +1000)]
[PATCH] m68knommu: need pfn_valid macro

Need pfn_valid macro, even on MMUless platforms.
Enclose the macro args of __pa and __va in parentheses.

Signed-off-by: Greg Ungerer <gerg@uclinux.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] m68knommu: use THREAD_SIZE instead of hard coded size
Greg Ungerer [Fri, 2 Sep 2005 00:42:52 +0000 (10:42 +1000)]
[PATCH] m68knommu: use THREAD_SIZE instead of hard coded size

Use the THREAD_SIZE define when manipulating the stack instead of
hard coded values (for the 68328 and 68360 sub-architectures).

Signed-off-by: Greg Ungerer <gerg@uclinux.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] m68knommu: new family (523x) and board config support
Greg Ungerer [Fri, 2 Sep 2005 00:42:52 +0000 (10:42 +1000)]
[PATCH] m68knommu: new family (523x) and board config support

New architecture and board configuration support for m68knommu.

. add 523x ColdFire support
. add support for SOM5282 and MOD5272 boards
. break up the 527x to be separate 5271 and 5275. There is some
  subtle differences that (like RAM config) that need to be dealt with
. add option to support selecting 4k kernel stack

Signed-off-by: Greg Ungerer <gerg@uclinux.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] m68knommu: 523x ColdFire processor support in arch Makefile
Greg Ungerer [Fri, 2 Sep 2005 00:42:52 +0000 (10:42 +1000)]
[PATCH] m68knommu: 523x ColdFire processor support in arch Makefile

Add support for the 523x ColdFire family of processors

Signed-off-by: Greg Ungerer <gerg@uclinux.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] m68knommu: cleanup showstack()
Greg Ungerer [Fri, 2 Sep 2005 00:42:52 +0000 (10:42 +1000)]
[PATCH] m68knommu: cleanup showstack()

Make show_stack() consistent with other architectures.
Put the vector string names in the .rodata section.
Patch originally submitted by Philippe De Muyter <phdm@macqel.be>.

Signed-off-by: Greg Ungerer <gerg@uclinux.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] uclinux: update MAINTAINERS entry for UCLINUX
Greg Ungerer [Fri, 2 Sep 2005 00:42:52 +0000 (10:42 +1000)]
[PATCH] uclinux: update MAINTAINERS entry for UCLINUX

Modify maintainers for uClinux (MMUless). Neither Dave nor Jeff
manitain the 2.6 code in mainline, so no point emailing them about
problems.

Signed-off-by: Greg Ungerer <gerg@uclinux.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] m68knommu: fix ColdFire startup code to properly handle non 0 based ram
Greg Ungerer [Fri, 2 Sep 2005 00:42:52 +0000 (10:42 +1000)]
[PATCH] m68knommu: fix ColdFire startup code to properly handle non 0 based ram

Correctly determine the end of ram for ram setups that do not
start at base address of 0. Add support for the MOD5272 board,
which doesn not have a ram base of 0.

Signed-off-by: Greg Ungerer <gerg@uclinux.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] m68knommu: new family (523x) and board setup
Greg Ungerer [Fri, 2 Sep 2005 00:42:52 +0000 (10:42 +1000)]
[PATCH] m68knommu: new family (523x) and board setup

. setup for the new 523x ColdFire family
. break up of 527x to be 5271 and 5275
. some white space cleanup

Signed-off-by: Greg Ungerer <gerg@uclinux.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] m68knommu: 523x ColdFire processor init code
Greg Ungerer [Fri, 2 Sep 2005 00:42:52 +0000 (10:42 +1000)]
[PATCH] m68knommu: 523x ColdFire processor init code

Low level initialization code for the 523x ColdFire processor family.

Signed-off-by: Greg Ungerer <gerg@uclinux.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge HEAD from master.kernel.org:/home/rmk/linux-2.6-serial
Linus Torvalds [Fri, 2 Sep 2005 07:53:36 +0000 (00:53 -0700)]
Merge HEAD from /home/rmk/linux-2.6-serial

19 years agoMerge HEAD from master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Fri, 2 Sep 2005 07:52:05 +0000 (00:52 -0700)]
Merge HEAD from kernel.org:/home/rmk/linux-2.6-arm

19 years agoMerge refs/heads/ieee80211-wifi from master.kernel.org:/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 2 Sep 2005 07:48:33 +0000 (00:48 -0700)]
Merge refs/heads/ieee80211-wifi from /linux/kernel/git/jgarzik/netdev-2.6

19 years agoMerge refs/heads/upstream from master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Fri, 2 Sep 2005 07:46:53 +0000 (00:46 -0700)]
Merge refs/heads/upstream from /linux/kernel/git/jgarzik/netdev-2.6

19 years ago/spare/repo/netdev-2.6 branch 'ieee80211'
Jeff Garzik [Thu, 1 Sep 2005 22:02:27 +0000 (18:02 -0400)]
/spare/repo/netdev-2.6 branch 'ieee80211'

19 years ago/spare/repo/netdev-2.6 branch 'master'
Jeff Garzik [Thu, 1 Sep 2005 22:02:01 +0000 (18:02 -0400)]
/spare/repo/netdev-2.6 branch 'master'

19 years ago[ARM] Fix ARMv6 page table bits
Russell King [Thu, 1 Sep 2005 21:41:55 +0000 (22:41 +0100)]
[ARM] Fix ARMv6 page table bits

We weren't explicitly setting the page table bits we desired
in user_prot in the protection table, which resulted in the
user mappings for v6 CPUs being marked global.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years agoMerge refs/heads/release from master.kernel.org:/pub/scm/linux/kernel/git/aegl/linux-2.6
Linus Torvalds [Thu, 1 Sep 2005 17:58:18 +0000 (10:58 -0700)]
Merge refs/heads/release from /linux/kernel/git/aegl/linux-2.6

19 years agoMerge HEAD from master.kernel.org:/home/rmk/linux-2.6-arm.git
Linus Torvalds [Thu, 1 Sep 2005 17:56:57 +0000 (10:56 -0700)]
Merge HEAD from /home/rmk/linux-2.6-arm.git

19 years ago[PATCH] ppc: L2 cache prefetch fixes on 745x
Kumar Gala [Wed, 31 Aug 2005 04:54:47 +0000 (14:54 +1000)]
[PATCH] ppc: L2 cache prefetch fixes on 745x

We run into problems if we blindly enable L2 prefetching without
checking that the L2 cache is actually enabled.  Additionaly, if we
disable the L2 cache we need to ensure that we disable L2 prefetching.

Signed-off-by: Kumar Gala <kumar.gala@freescale.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] Fix PCI ROM mapping
Benjamin Herrenschmidt [Wed, 31 Aug 2005 04:16:53 +0000 (14:16 +1000)]
[PATCH] Fix PCI ROM mapping

This fixes a problem with pci_map_rom() which doesn't properly
update the ROM BAR value with the address thas allocated for it by the
PCI code. This problem, among other, breaks boot on Mac laptops.

It'ss a new version based on Linus latest one with better error
checking.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] Fix bug in ppc64 dynamic hugepage support
David Gibson [Wed, 31 Aug 2005 04:34:05 +0000 (14:34 +1000)]
[PATCH] Fix bug in ppc64 dynamic hugepage support

In adjusting the logic for SLB miss for the dynamic hugepage stuff, I
messed up the !CONFIG_HUGETLB_PAGE case, failing to set the SLB flags
properly.

This fixes it.  It also streamlines the logic for the HUGETLB_PAGE case
(removing a couple of branches) while we're at it.

Booted, and roughly tested on POWER5 (with and without HUGETLB_PAGE),
iSeries/RS64 (no hugepage available), and G5 (with and without
HUGETLB_PAGE).

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] Add missing select's to DVB_BUDGET_AV
Adrian Bunk [Wed, 31 Aug 2005 15:43:51 +0000 (17:43 +0200)]
[PATCH] Add missing select's to DVB_BUDGET_AV

This fixes the following compile error:

...
  LD      .tmp_vmlinux1
drivers/built-in.o: In function `frontend_init':
budget-av.c:(.text+0xb9448): undefined reference to `tda10046_attach'
budget-av.c:(.text+0xb9518): undefined reference to `tda10021_attach'
drivers/built-in.o: In function `philips_tu1216_request_firmware':
budget-av.c:(.text+0xb937b): undefined reference to `request_firmware'
make: *** [.tmp_vmlinux1] Error 1

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Acked-by: Johannes Stezenbach <js@linuxtv.org>
Acked-by: Michael Krufky <mkrufky@linuxtv.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[SERIAL] Move serial8250_*_port prototypes to linux/serial_8250.h
Russell King [Thu, 1 Sep 2005 14:56:26 +0000 (15:56 +0100)]
[SERIAL] Move serial8250_*_port prototypes to linux/serial_8250.h

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] Simplify setup_mm_for_reboot()
Russell King [Thu, 1 Sep 2005 13:51:59 +0000 (14:51 +0100)]
[ARM] Simplify setup_mm_for_reboot()

No point checking what CPU architecture level we have each time
within the loop, so precompute the base PMD flags outside the
loop.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] Convert open-coded __pmd_populate to use inline function
Russell King [Thu, 1 Sep 2005 13:45:18 +0000 (14:45 +0100)]
[ARM] Convert open-coded __pmd_populate to use inline function

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[SERIAL] mwave is no longer broken
Russell King [Thu, 1 Sep 2005 13:25:45 +0000 (14:25 +0100)]
[SERIAL] mwave is no longer broken

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] 2864/1: VST aka CONFIG_NO_IDLE_HZ support for SA11x0
Nicolas Pitre [Thu, 1 Sep 2005 11:48:48 +0000 (12:48 +0100)]
[ARM] 2864/1: VST aka CONFIG_NO_IDLE_HZ support for SA11x0

Patch from Nicolas Pitre

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] 2863/1: clarify comment in PXA2xx and SA1x00 timer code
Nicolas Pitre [Thu, 1 Sep 2005 11:48:47 +0000 (12:48 +0100)]
[ARM] 2863/1: clarify comment in PXA2xx and SA1x00 timer code

Patch from Nicolas Pitre

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] 2862/1: VST aka CONFIG_NO_IDLE_HZ support for PXA2xx
Nicolas Pitre [Thu, 1 Sep 2005 11:48:40 +0000 (12:48 +0100)]
[ARM] 2862/1: VST aka CONFIG_NO_IDLE_HZ support for PXA2xx

Patch from Nicolas Pitre

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] 2865/2: fix fadvise64_64 syscall argument passing
Nicolas Pitre [Thu, 1 Sep 2005 11:37:13 +0000 (12:37 +0100)]
[ARM] 2865/2: fix fadvise64_64 syscall argument passing

Patch from Nicolas Pitre

The prototype for sys_fadvise64_64() is:
    long sys_fadvise64_64(int fd, loff_t offset, loff_t len, int advice)
The argument list is therefore as follows on legacy ABI:
fd: type int (r0)
offset: type long long (r1-r2)
len: type long long (r3-sp[0])
advice: type int (sp[4])
With EABI this becomes:
fd: type int (r0)
offset: type long long (r2-r3)
len: type long long (sp[0]-sp[4])
advice: type int (sp[8])
Not only do we have ABI differences here, but the EABI version requires
one additional word on the syscall stack.
To avoid the ABI mismatch and the extra stack space required with EABI
this syscall is now defined with a different argument ordering
on ARM as follows:
    long sys_arm_fadvise64_64(int fd, int advice, loff_t offset, loff_t len)
This gives us the following ABI independent argument distribution:
fd: type int (r0)
advice: type int (r1)
offset: type long long (r2-r3)
len: type long long (sp[0]-sp[4])
Now, since the syscall entry code takes care of 5 registers only by
default including the store of r4 to the stack, we need a wrapper to
store r5 to the stack as well.  Because that wrapper was missing and was
always required this means that sys_fadvise64_64 never worked on ARM and
therefore we can safely reuse its syscall number for our new
sys_arm_fadvise64_64 interface.

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[PATCH] hostap: Fix null pointer dereference in prism2_pccard_card_present()
Jouni Malinen [Mon, 29 Aug 2005 00:53:32 +0000 (17:53 -0700)]
[PATCH] hostap: Fix null pointer dereference in prism2_pccard_card_present()

local->hw_priv was initialized only after the interrupt handler was
registered. This could trigger a NULL pointer dereference in
prism2_pccard_card_present() that assumed that local->hw_priv is always
set (and it should have been). Fix this by setting local->hw_priv before
registering the interrupt handler.

Signed-off-by: Jouni Malinen <jkmaline@cc.hut.fi>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Be consistent about driver name, increment version
Michael Ellerman [Thu, 1 Sep 2005 01:29:29 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Be consistent about driver name, increment version

The iseries_veth driver tells sysfs that it's called 'iseries_veth', but if
you ask it via ethtool it thinks it's called 'veth'. I think this comes from
2.4 when the driver was called 'veth', but it's definitely called
'iseries_veth' now, so fix it.

To make sure we don't do it again define DRV_NAME and use it everywhere.

While we're at it, change the version number to 2.0, to reflect the changes
made in this patch series.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Remove studly caps from iseries_veth.c
Michael Ellerman [Thu, 1 Sep 2005 01:29:27 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Remove studly caps from iseries_veth.c

Having merged iseries_veth.h, let's remove some of the studly caps that came
with it.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Incorporate iseries_veth.h in iseries_veth.c
Michael Ellerman [Thu, 1 Sep 2005 01:29:25 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Incorporate iseries_veth.h in iseries_veth.c

iseries_veth.h is only used by iseries_veth.c, so merge the former into
the latter.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Add sysfs support for port structs
Michael Ellerman [Thu, 1 Sep 2005 01:29:21 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Add sysfs support for port structs

Also to aid debugging, add sysfs support for iseries_veth's port structures.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Add sysfs support for connection structs
Michael Ellerman [Thu, 1 Sep 2005 01:29:20 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Add sysfs support for connection structs

To aid in field debugging, add sysfs support for iseries_veth's connection
structures. At the moment this is all read-only, however we could think about
adding write support for some attributes in future.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Fix bogus counting of TX errors
Michael Ellerman [Thu, 1 Sep 2005 01:29:19 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Fix bogus counting of TX errors

There's a number of problems with the way iseries_veth counts TX errors.

Firstly it counts conditions which aren't really errors as TX errors. This
includes if we don't have a connection struct for the other LPAR, or if the
other LPAR is currently down (or just doesn't want to talk to us). Neither
of these should count as TX errors.

Secondly, it counts one TX error for each LPAR that fails to accept the packet.
This can lead to TX error counts higher than the total number of packets sent
through the interface. This is confusing for users.

This patch fixes that behaviour. The non-error conditions are no longer
counted, and we introduce a new and I think saner meaning to the TX counts.

If a packet is successfully transmitted to any LPAR then it is transmitted
and tx_packets is incremented by 1.

If there is an error transmitting a packet to any LPAR then that is counted
as one error, ie. tx_errors is incremented by 1.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Simplify full-queue handling
Michael Ellerman [Thu, 1 Sep 2005 01:29:18 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Simplify full-queue handling

The iseries_veth driver often has multiple netdevices sending packets over
a single connection to another LPAR. If the bandwidth to the other LPAR is
exceeded, all the netdevices must have their queues stopped.

The current code achieves this by queueing one incoming skb on the
per-netdevice port structure. When the connection is able to send more packets
we iterate through the port structs and flush any packet that is queued,
as well as restarting the associated netdevice's queue.

This arrangement makes less sense now that we have per-connection TX timers,
rather than the per-netdevice generic TX timer.

The new code simply detects when one of the connections is full, and stops
the queue of all associated netdevices. Then when a packet is acked on that
connection (ie. there is space again) all the queues are woken up.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Add a per-connection ack timer
Michael Ellerman [Thu, 1 Sep 2005 01:29:17 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Add a per-connection ack timer

Currently the iseries_veth driver contravenes the specification in
Documentation/networking/driver.txt, in that if packets are not acked by
the other LPAR they will sit around forever.

This patch adds a per-connection timer which fires if we've had no acks for
five seconds. This is superior to the generic TX timer because it catches
the case of a small number of packets being sent and never acked.

This fixes a bug we were seeing on real systems, where some IPv6 neighbour
discovery packets would not be acked and then prevent the module from being
removed, due to skbs lying around.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Remove TX timeout code
Michael Ellerman [Thu, 1 Sep 2005 01:29:12 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Remove TX timeout code

The iseries_veth driver uses the generic TX timeout watchdog, however a better
solution is in the works, so remove this code.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Use kobjects to track lifecycle of connection structs
Michael Ellerman [Thu, 1 Sep 2005 01:29:09 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Use kobjects to track lifecycle of connection structs

The iseries_veth driver can attach to multiple vlans, which correspond to
multiple net devices. However there is only 1 connection between each LPAR,
so the connection structure may be shared by multiple net devices.

This makes module removal messy, because we can't deallocate the connections
until we know there are no net devices still using them. The solution is to
use ref counts on the connections, so we can delete them (actually stop) as
soon as the ref count hits zero.

This patch fixes (part of) a bug we were seeing with IPv6 sending probes to
a dead LPAR, which would then hang us forever due to leftover skbs.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Make init_connection() & destroy_connection() symmetrical
Michael Ellerman [Thu, 1 Sep 2005 01:29:08 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Make init_connection() & destroy_connection() symmetrical

This patch makes veth_init_connection() and veth_destroy_connection()
symmetrical in that they allocate/deallocate the same data.

Currently if there's an error while initialising connections (ie. ENOMEM)
we call veth_module_cleanup(), however this will oops because we call
driver_unregister() before we've called driver_register(). I've never seen
this actually happen though.

So instead we explicitly call veth_destroy_connection() for each connection,
any that have been set up will be deallocated.

We also fix a potential leak if vio_register_driver() fails.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Only call dma_unmap_single() if dma_map_single() succeeded
Michael Ellerman [Thu, 1 Sep 2005 01:29:07 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Only call dma_unmap_single() if dma_map_single() succeeded

The iseries_veth driver unconditionally calls dma_unmap_single() even
when the corresponding dma_map_single() may have failed.

Rework the code a bit to keep the return value from dma_unmap_single()
around, and then check if it's a dma_mapping_error() before we do
the dma_unmap_single().

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Replace lock-protected atomic with an ordinary variable
Michael Ellerman [Thu, 1 Sep 2005 01:29:06 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Replace lock-protected atomic with an ordinary variable

The iseries_veth driver uses atomic ops to manipulate the in_use field of
one of its per-connection structures. However all references to the
flag occur while the connection's lock is held, so the atomic ops aren't
necessary.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Remove redundant message stack lock
Michael Ellerman [Thu, 1 Sep 2005 01:29:05 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Remove redundant message stack lock

The iseries_veth driver keeps a stack of messages for each connection
and a lock to protect the stack. However there is also a per-connection lock
which makes the message stack lock redundant.

Remove the message stack lock and document the fact that callers of the
stack-manipulation functions must hold the connection's lock.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Fix broken promiscuous handling
Michael Ellerman [Thu, 1 Sep 2005 01:29:02 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Fix broken promiscuous handling

Due to a logic bug, once promiscuous mode is enabled in the iseries_veth
driver it is never disabled.

The driver keeps two flags, promiscuous and all_mcast which have exactly the
same effect. This is because we only ever receive packets destined for us,
or multicast packets. So consolidate them into one promiscuous flag for
simplicity.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Try to avoid pathological reset behaviour
Michael Ellerman [Thu, 1 Sep 2005 01:29:00 +0000 (11:29 +1000)]
[PATCH] iseries_veth: Try to avoid pathological reset behaviour

The iseries_veth driver contains a state machine which is used to manage
how connections are setup and neogotiated between LPARs.

If one side of a connection resets for some reason, the two LPARs can get
stuck in a race to re-setup the connection. This can lead to the connection
being declared dead by one or both ends. In practice the connection is
declared dead by one or both ends approximately 8/10 times a connection is
reset, although it is rare for connections to be reset.

(an example here: http://michael.ellerman.id.au/files/misc/veth-trace.html)

The core of the problem is that the end that resets the connection doesn't
wait for the other end to become aware of the reset. So the resetting end
starts setting the connection back up, and then receives a reset from the
other end (which is the response to the initial reset). And so on.

We're severely limited in what we can do to fix this. The protocol between
LPARs is essentially fixed, as we have to interoperate with both OS/400
and old Linux drivers. Which also means we need a fix that only changes the
code on one end.

The only fix I've found given that, is to just blindly sleep for a bit when
resetting the connection, in the hope that the other end will get itself
sorted.  Needless to say I'd love it if someone has a better idea.

This does work, I've so far been unable to get it to break, whereas without
the fix a reset of one end will lead to a dead connection ~8/10 times.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Remove a FIXME WRT deletion of the ack_timer
Michael Ellerman [Thu, 1 Sep 2005 01:28:59 +0000 (11:28 +1000)]
[PATCH] iseries_veth: Remove a FIXME WRT deletion of the ack_timer

The iseries_veth driver has a timer which we use to send acks. When the
connection is reset or stopped we need to delete the timer.

Currently we only call del_timer() when resetting a connection, which means
the timer might run again while the connection is being re-setup. As it turns
out that's ok, because the flags the timer consults have been reset.

It's cleaner though to call del_timer_sync() once we've dropped the lock,
although the timer may still run between us dropping the lock and calling
del_timer_sync(), but as above that's ok.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] iseries_veth: Cleanup error and debug messages
Michael Ellerman [Thu, 1 Sep 2005 01:28:57 +0000 (11:28 +1000)]
[PATCH] iseries_veth: Cleanup error and debug messages

Currently the iseries_veth driver prints the file name and line number in its
error messages. This isn't very useful for most users, so just print
"iseries_veth: message" instead.

 - convert uses of veth_printk() to veth_debug()/veth_error()/veth_info()
 - make terminology consistent, ie. always refer to LPAR not lpar
 - be consistent about printing return codes as %d not %x
 - make format strings fit in 80 columns

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years agoAuto-update from upstream
Tony Luck [Wed, 31 Aug 2005 21:19:44 +0000 (14:19 -0700)]
Auto-update from upstream

19 years ago[SERIAL] Convert mwave to use serial8250_(un)?register_port
Alan Cox [Wed, 31 Aug 2005 21:19:33 +0000 (22:19 +0100)]
[SERIAL] Convert mwave to use serial8250_(un)?register_port

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[IA64] uncached allocator: use generic (not sn2 specific) functions
Martin Hicks [Wed, 31 Aug 2005 18:21:00 +0000 (11:21 -0700)]
[IA64] uncached allocator: use generic (not sn2 specific) functions

Change sn2-specific calls into generic functions.  Without this change
the uncached allocator will not work on non-sn2 platforms.

Signed-off-by: Greg Edwards <edwardsg@sgi.com>
Signed-off-by: Martin Hicks <mort@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[ARM] 2866/1: add i.MX set_mctrl / get_mctrl functions
Sascha Hauer [Wed, 31 Aug 2005 20:48:47 +0000 (21:48 +0100)]
[ARM] 2866/1: add i.MX set_mctrl / get_mctrl functions

Patch from Sascha Hauer

This patch adds support for setting and getting RTS / CTS via
set_mtctrl / get_mctrl functions.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] 2869/1: ixp4xx: correct ioread*/iowrite*
David Vrabel [Wed, 31 Aug 2005 20:45:14 +0000 (21:45 +0100)]
[ARM] 2869/1: ixp4xx: correct ioread*/iowrite*

Patch from David Vrabel

Correct the ioread* and iowrite* functions.  In particular, add an offset to the cookie in ioport_map so we can map I/O port ranges starting from 0 (0 is for reporting errors).

Signed-off-by: David Vrabel <dvrabel@arcom.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] 2868/1: Include linux/cpumask.h in arch/arm/common/gic.c
Catalin Marinas [Wed, 31 Aug 2005 20:45:14 +0000 (21:45 +0100)]
[ARM] 2868/1: Include linux/cpumask.h in arch/arm/common/gic.c

Patch from Catalin Marinas

Minor compilation error fix.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] 2867/2: unaligned ldrd/strd fixups
Steve Longerbeam [Wed, 31 Aug 2005 20:22:20 +0000 (21:22 +0100)]
[ARM] 2867/2: unaligned ldrd/strd fixups

Patch from Steve Longerbeam

Adds an implementation of unaligned LDRD and STRD fixups.
Also fixes a bug where do_alignment() would misinterpret and
fixup an unaligned LDRD/STRD as LDRH/STRH, causing memory
corruption.
This is the same as Patch #2867/1, but with minor whitespace
and comments changes, plus a check for arch-level >= v5TE
before printing ai_dword count in proc_alignment_read().

Signed-off-by: Steve Longerbeam <stevel@mwwireless.net>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[netdrvr de2104x] store PCI bus addresses in unsigned long
Jeff Garzik [Wed, 31 Aug 2005 10:11:16 +0000 (06:11 -0400)]
[netdrvr de2104x] store PCI bus addresses in unsigned long

BZ# 4475.

19 years ago[netdrvr tulip] new PCI ID
Jeff Garzik [Wed, 31 Aug 2005 09:48:59 +0000 (05:48 -0400)]
[netdrvr tulip] new PCI ID

Noted in BZ# 2960.

19 years ago[SERIAL] Clean up and fix tty transmission start/stoping
Russell King [Wed, 31 Aug 2005 09:12:14 +0000 (10:12 +0100)]
[SERIAL] Clean up and fix tty transmission start/stoping

The start_tx and stop_tx methods were passed a flag to indicate
whether the start/stop was from the tty start/stop callbacks, and
some drivers used this flag to decide whether to ask the UART to
immediately stop transmission (where the UART supports such a
feature.)

There are other cases when we wish this to occur - when CTS is
lowered, or if we change from soft to hard flow control and CTS
is inactive.  In these cases, this flag was false, and we would
allow the transmitter to drain before stopping.

There is really only one case where we want to let the transmitter
drain before disabling, and that's when we run out of characters
to send.

Hence, re-jig the start_tx and stop_tx methods to eliminate this
flag, and introduce new functions for the special "disable and
allow transmitter to drain" case.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] 2857/2: Dynamic tick - fix OOPS if configured and not provided
Ben Dooks [Wed, 31 Aug 2005 08:42:14 +0000 (09:42 +0100)]
[ARM] 2857/2: Dynamic tick - fix OOPS if configured and not provided

Patch from Ben Dooks

timer_dyn_reprogram() fails with an OOPS if the
configuration for CONFIG_NO_IDLE_HZ is enabled, and
the system has no support for it.

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[IA64] Low byte of current->personality is not a bitmask.
Tony Luck [Tue, 30 Aug 2005 21:59:24 +0000 (14:59 -0700)]
[IA64] Low byte of current->personality is not a bitmask.

Peter Staubach pointed out that it is not correct to check
current->personality & PER_LINUX32 (this will have false
hits on several other personality values).

Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years agoMerge refs/heads/upstream from master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Tue, 30 Aug 2005 18:16:30 +0000 (11:16 -0700)]
Merge refs/heads/upstream from /linux/kernel/git/jgarzik/netdev-2.6.git

19 years ago[PATCH] ppc32 8xx: fix warnings in m8xx_setup.c
Marcelo Tosatti [Tue, 30 Aug 2005 16:40:22 +0000 (13:40 -0300)]
[PATCH] ppc32 8xx: fix warnings in m8xx_setup.c

The following patch fixes two warnings in arch/ppc/syslib/m8xx_setup.c

Signed-off-by: Marcelo Tosatti <marcelo.tosatti@cyclades.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] x86: pci_assign_unassigned_resources() update
Ivan Kokshaysky [Tue, 30 Aug 2005 14:48:52 +0000 (18:48 +0400)]
[PATCH] x86: pci_assign_unassigned_resources() update

I had some time to think about PCI assign issues in 2.6.13-rc series.

The major problem here is that we call pci_assign_unassigned_resources()
way too early - at subsys_initcall level. Therefore we give no chances
to ACPI and PnP routines (called at fs_initcall level) to reserve their
respective resources properly, as the comments in drivers/pnp/system.c
and drivers/acpi/motherboard.c suggest:

 /**
  * Reserve motherboard resources after PCI claim BARs,
  * but before PCI assign resources for uninitialized PCI devices
  */

So I moved the pci_assign_unassigned_resources() call to
pcibios_assign_resources() (fs_initcall), which should hopefully fix a
lot of problems and make PCIBIOS_MIN_IO tweaks unnecessary.

Other changes:
- remove resource assignment code from pcibios_assign_resources(), since
  it duplicates pci_assign_unassigned_resources() functionality and
  actually does nothing in 2.6.13;
- modify ROM assignment code as per Ben's suggestion: try to use firmware
  settings by default (if PCI_ASSIGN_ROMS is not set);
- set CARDBUS_IO_SIZE back to 4K as it's a wonderful stress test for
  various setups.

Confirmed by Tero Roponen <teanropo@cc.jyu.fi> (who had problems with
the 4kB CardBus IO size previously).

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge HEAD from /spare/repo/linux-2.6/.git
Jeff Garzik [Tue, 30 Aug 2005 17:32:29 +0000 (13:32 -0400)]
Merge HEAD from /spare/repo/linux-2.6/.git

19 years agoAuto-update from upstream
Tony Luck [Tue, 30 Aug 2005 16:30:09 +0000 (09:30 -0700)]
Auto-update from upstream

19 years agoMerge refs/heads/for-linus from master.kernel.org:/pub/scm/linux/kernel/git/shaggy...
Linus Torvalds [Tue, 30 Aug 2005 14:47:42 +0000 (07:47 -0700)]
Merge refs/heads/for-linus from /linux/kernel/git/shaggy/jfs-2.6.git

19 years agoMerge HEAD from master.kernel.org:/pub/scm/linux/kernel/git/perex/alsa
Linus Torvalds [Tue, 30 Aug 2005 14:47:01 +0000 (07:47 -0700)]
Merge HEAD from /linux/kernel/git/perex/alsa

19 years agoMerge refs/heads/drm-latest from master.kernel.org:/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Tue, 30 Aug 2005 14:45:15 +0000 (07:45 -0700)]
Merge refs/heads/drm-latest from /linux/kernel/git/airlied/drm-2.6

19 years ago[ALSA] usb-audio: throttle MIDI URB resubmits on USB errors
Clemens Ladisch [Tue, 2 Aug 2005 13:26:52 +0000 (15:26 +0200)]
[ALSA] usb-audio: throttle MIDI URB resubmits on USB errors

USB generic driver
When a USB error occurs that might indicate that the device has been
unplugged, don't resubmit the URB immediately to prevent flooding the
log with error messages before khubd has us disconnect()ed.

Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
19 years ago[PATCH] Assign device pointer to OSS devices
Takashi Iwai [Tue, 30 Aug 2005 06:58:37 +0000 (08:58 +0200)]
[PATCH] Assign device pointer to OSS devices

Add register_sound_special_device() function to allow assignment of
device pointer to a specific OSS device for HAL.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
19 years ago[ALSA] version 1.0.10rc1
Jaroslav Kysela [Tue, 30 Aug 2005 06:48:35 +0000 (08:48 +0200)]
[ALSA] version 1.0.10rc1

19 years ago[ALSA] Add identification for Live 5.1 [SB0220]
Lee Revell [Mon, 29 Aug 2005 15:42:00 +0000 (17:42 +0200)]
[ALSA] Add identification for Live 5.1 [SB0220]

EMU10K1/EMU10K2 driver
Add identification for Live 5.1 [SB0220]

Signed-off-by: Lee Revell <rlrevell@joe-job.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
19 years ago[ALSA] emu10k1 - Add missing ac97 support on SBLive! Player 5.1
Takashi Iwai [Fri, 26 Aug 2005 15:26:40 +0000 (17:26 +0200)]
[ALSA] emu10k1 - Add missing ac97 support on SBLive! Player 5.1

EMU10K1/EMU10K2 driver
Added the missing ac97 support on SBLive! Player 5.1.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
19 years ago[ALSA] intel8x0 - Add quirk for FSC 4010
Takashi Iwai [Thu, 25 Aug 2005 18:30:27 +0000 (20:30 +0200)]
[ALSA] intel8x0 - Add quirk for FSC 4010

Intel8x0 driver
Added ac97_quirk for FSC 4010.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
19 years ago[ALSA] pcm - Fix zero-division in 32bit compat layer
Takashi Iwai [Thu, 25 Aug 2005 17:51:47 +0000 (19:51 +0200)]
[ALSA] pcm - Fix zero-division in 32bit compat layer

PCM Midlevel
Fixed zero-division bug in PCM 32bit compat layer.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
19 years ago[ALSA] intel8x0 - add AC97_TUNE_MUTE_LED quirk for HP nx8220 laptop 103c:0934
Daniel Horchner [Thu, 25 Aug 2005 11:02:24 +0000 (13:02 +0200)]
[ALSA] intel8x0 - add AC97_TUNE_MUTE_LED quirk for HP nx8220 laptop 103c:0934

Intel8x0 driver

Signed-off-by: Jaroslav Kysela <perex@suse.cz>
19 years ago[ALSA] usb-audio: fix Emagic MIDI protocol handling
Clemens Ladisch [Thu, 25 Aug 2005 09:10:05 +0000 (11:10 +0200)]
[ALSA] usb-audio: fix Emagic MIDI protocol handling

USB generic driver
Emagic devices pad their packets not with 0xff bytes but with a 0xff
byte followed by garbage, so we have to stop at the first such byte.

Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
19 years ago[ALSA] Fix missing spin_unlock
Takashi Iwai [Wed, 24 Aug 2005 15:57:25 +0000 (17:57 +0200)]
[ALSA] Fix missing spin_unlock

au88x0 driver,Common EMU synth
Fixed missing spin_unlock.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
19 years ago[ALSA] include/sound/gus.h: 'extern inline' -> 'static inline'
Adrian Bunk [Wed, 24 Aug 2005 15:01:02 +0000 (17:01 +0200)]
[ALSA] include/sound/gus.h: 'extern inline' -> 'static inline'

GUS Library
'extern inline' doesn't make much sense.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
19 years ago[ALSA] hda-intel - Fix ULI M5461 support
Takashi Iwai [Wed, 24 Aug 2005 12:14:57 +0000 (14:14 +0200)]
[ALSA] hda-intel - Fix ULI M5461 support

HDA Intel driver
Fix and clean up for the support of ULI M5461
- set CORB/RIRB sizes explicitly
- add workarounds for ULI on ia32
- max number of streams depends on the chip type now
- increase the size of BDL

Signed-off-by: Takashi Iwai <tiwai@suse.de>