Bill Nottingham [Thu, 22 Sep 2005 07:47:36 +0000 (00:47 -0700)]
[PATCH] fix class symlinks in sysfs
The class symlinks in sysfs don't properly handle changing device names.
To demonstrate, rename your network device from eth0 to eth1. Your
pci (or usb, or whatever) device will still have a 'net:eth0' link,
except now it points to /sys/class/net/eth1.
The attached patch makes sure the class symlink name changes when
the class device name changes. It isn't 100% correct, it should be
using sysfs_rename_link. Unfortunately, sysfs_rename_link doesn't exist.
Daniel Ritz [Thu, 22 Sep 2005 07:47:11 +0000 (00:47 -0700)]
[PATCH] Driver Core: fis bus rescan devices race
bus_rescan_devices_helper() does not hold the dev->sem when it checks for
!dev->driver(). device_attach() holds the sem, but calls again
device_bind_driver() even when dev->driver is set.
What happens is that a first device_attach() call (module insertion time)
is on the way binding the device to a driver. Another thread calls
bus_rescan_devices(). Now when bus_rescan_devices_helper() checks for
dev->driver it is still NULL 'cos the the prior device_attach() is not yet
finished. But as soon as the first one releases the dev->sem the second
device_attach() tries to rebind the already bound device again.
device_bind_driver() does this blindly which leads to a corrupt
driver->klist_devices list (the device links itself, the head points to the
device). Later a call to device_release_driver() sets dev->driver to NULL
and breaks the link it has to itself on knode_driver. Rmmoding the driver
later calls driver_detach() which leads to an endless loop 'cos the list
head in klist_devices still points to the device. And since dev->driver is
NULL it's stuck with the same device forever. Boom. And rmmod hangs.
Very easy to reproduce with new-style pcmcia and a 16bit card. Just loop
modprobe <pcmcia-modules> ;cardctl eject; rmmod <card driver, pcmcia
modules>.
Easiest fix is to check if the device is already bound to a driver in
device_bind_driver(). This avoids the double binding.
Kumar Gala [Wed, 21 Sep 2005 21:52:55 +0000 (16:52 -0500)]
[PATCH] powerpc: merged hw_irq.h
Merged hw_irq.h between ppc32 & ppc64. Added support to use the Book-E
wrtee[i] instructions that allow modifying MSR[EE] atomically.
Additionally, added get_irq_desc() macros to ppc32 to allow mask_irq(),
unmask_irq(), and ack_irq() to be common between ppc32 & ppc64.
Note: because 64-bit Book-E implementations only have a 32-bit MSR the
macro's for Book-E need to come before the PPC64 macro's to ensure the
right thing happends for 64-bit Book-E processors.
Kumar Gala [Wed, 21 Sep 2005 20:44:11 +0000 (15:44 -0500)]
[PATCH] powerpc: Fix compiling of ppc32
The merging of auxvec.h into asm-powerpc introduced the AT_SYSINFO_EHDR
into the ppc32 build that is used for VDSO. However, we dont have VDSO
support in the ppc32 tree at this time. Introducing this define causes
a number of other things to get built with the assumption of VDSO, thus
causing the compile errors for ppc32.
Until we have VDSO on ppc32 we will leave AT_SYSINFO_EHDR a ppc64 only
define.
Cleanup receive buffer allocation and management,
Add more error handling checks from PHY and bump version. Signed-off-by: Jeff Garzik <[email protected]>
Expand the returned data for ethtool debug access to include
all of the mapped PCI area; except for the small set of registers
that are for diagnostic RAM access. Access to those registers
will hang the system.
[PATCH] bonding: Fix link monitor capability check (was skge: set mac address oops with bonding)
Fix bond_enslave link monitoring warning to check use_carrier status
and ethtool_ops in addition to do_ioctl. This version checks ethtool_ops
as well as do_ioctl, and also uses the per-bond params.use_carrier
instead of the global use_carrier.
[PATCH] uml: fix compile warning after consolidation patch
The header declaring this function wasn't included, so the function declaration
was totally bogus wrt. the proto - even if this wasn't going to fail at all.
It was so bad that the compile warning I got was "control reaches end of
non-void function", i.e. missing return. Actually, this has been there for ages,
the consolidation patch just added the warning which was needed to clean it up.
Nice. Really.
Well, the symbol was intended to come from userspace (it exists there on normal
host), but since some hosts may miss that, using the kernel one is just as fine.
However, rename it to be named consistently with the rest.
Actually, he missed converting ELFCLASS32 to coming from kernel headers. For
consistence, add ELFCLASS64 too.
[PATCH] uml: fix modify_ldt - missing break in switch
I am a lamer :-(. Luckily, Luo Xin performed LTP testing and found this failure.
Btw, the fact that the patch in which I introduced this was merged shows that:
a) I'm really trusted by people
b) sometimes they're wrong about point a).
c) lack of time for reviewers.
As recently done by Russell King for ARM, commit 4732efbeb997189d9f9b04708dc26bf8613ed721 introduces a generic asm/futex.h copied
along most arches, which includes a "-ENOSYS support" to be changed if needed.
However, it includes an unused var (taken from the "real" version) which GCC
warns about.
Remove it from all arches having that file version (i.e. same GIT id).
$ git-diff-tree -r HEAD
and
$ git-ls-tree -r HEAD include/|grep 9feff4ce1424bc390608326240be369eb13aa648
may be more interesting than looking at the patch itself, to make sure I've
just copied the arm header to all other archs having the original dummy version
of this file.
Follow up to 4732efbeb997189d9f9b04708dc26bf8613ed721 - uml must just reuse
as-is the backing architecture support. There is a micro-fixup is needed for the
included file, which won't affect i386 behaviour at all.
I've not tested compilation on x86_64, only on x86, but the code is almost the
same except the culprit test, so everything should be ok on x86_64 too.
During a forensic analysis on the fat file system, I found than the result for
the last access date on this file system was different between the stat
command and the istat command (package tct-utils).
The istat command display a true date (the right windows date) but the stat
primitive (so stat, find, ls command) displays a wrong date.
__FUNCTION__ is the prefered kernel idiom, __func__ is not supported by gcc
2.95 (we actually map __FUNCTION__ to __func__ for more recent compilers,
but it should never be used directly)
[PATCH] Fix I2O config-osm init to return proper error
We currently unregister the config-osm driver if initialization of the
legacy ioctl() handlers failed but still return success. We should be
returning -EBUSY in this case.
zImage.vmode was recently added. It's a version of zImage in which the ELF
note section used by open firmware indicates that it requires a virtual
mode instance of OF instead of real mode. This allows it to work with
Apple OF, and thus is directly bootable (or netbootable) from OF command
line. (Unfortunately, pSeries OF sort-of requires real mode and Apple OF
sort-of requires virtual mode, and both tend to be unhappy if no notes
section specifies the mode at all).
However, we forgot to add zImage.vmode to the default G5 build. This
fixes it.
Olof Johansson [Wed, 21 Sep 2005 16:55:31 +0000 (09:55 -0700)]
[PATCH] PPC64: Fix boot for some pre-POWER4 systems
Some RS64 systems (such as F80) have non-python host bridges with EADS.
However, they have two EADS with 4 buses each under them, so the old logic
that assumed no more than 7 busses per PHB failed miserably.
Big thanks to Olaf Hering for helping me test this, he's got one of the few
machines that broke from the previous logic.
Also, to be a bit smarter at detecting the need for a PHB-level IOMMU table
by checking for the presence of an ISA bus. Only PHBs with ISA bridges
should need the PHB-level table.
Andrew Morton [Wed, 21 Sep 2005 16:55:30 +0000 (09:55 -0700)]
[PATCH] Add printk_clock()
ia64's sched_clock() accesses per-cpu data which isn't set up at boot time.
Hence ia64 cannot use printk timestamping, because printk() will crash in
sched_clock().
So make printk() use printk_clock(), defaulting to sched_clock(), overrideable
by the architecture via attribute(weak).
When the main thread of a thread group has done pthread_exit() and died,
the other threads are still happily running, but will not be visible
under /proc because their leader is no longer accessible.
This fixes the access control so that we can see the sub-threads again.
Paul Mackerras [Wed, 21 Sep 2005 04:14:22 +0000 (14:14 +1000)]
[PATCH] ppc64: Fix PCI flags when using OF device tree
My code to set up the PCI tree from the Open Firmware device tree was
setting IORESOURCE_* flags on the resources for the devices, but not
the PCI_BASE_ADDRESS_* flags. This meant that some drivers
misbehaved, and /proc/pci showed the wrong types for the resources.
This fixes it.
Becky Bruce [Mon, 19 Sep 2005 22:01:54 +0000 (17:01 -0500)]
[PATCH] powerpc: Merge bug.h
ppc32/ppc64: Merge bug.h into include/asm-powerpc
This patch merges bug.h into include/asm-powerpc. Changed the data
structure for bug_entry such that line is always an int on both 32 and
64-bit platforms; removed casts to int from the 64-bit trap code to
reflect this.
This is a revised patch to merge asm-ppc*/hardirq.h.
It removes some unnecessary #includes, but then requires
the addition of #include <asm/irq.h> in PPC32's hw_irq.h
much like ppc64 already does. Furthermore, several
unnecessary #includes were removed from some ppc32 boards
in order to break resulting bad #include cycles.
Builds pSeries_defconfig and all ppc32 platforms except
the already b0rken bseip.
Kumar Gala [Mon, 19 Sep 2005 14:30:27 +0000 (09:30 -0500)]
[PATCH] powerpc: unified signature of timer_interrupt() between ppc32/ppc64
On ppc64 timer_interrupt() returned a value that was never used. Changed
the ppc64 version of timer_interrupt() to no longer return a value so
that the signatures between ppc32 & ppc64 match. This will simplify
future merging of arch/powerpc.
Kumar Gala [Mon, 19 Sep 2005 02:33:18 +0000 (21:33 -0500)]
[PATCH] ppc32: Allow user to individual select CHRP/PMAC/PREP config
Made the CHRP/PMAC/PREP config options selectable by the user.
This allows us to build kernels specifically for one of the
platforms thus reducing code size.
[PATCH] ppc64: Store virtual address in TLB flush batches
This patch slightly change the TLB flush batch mecanism so that we
store the full vaddr (including vsid) when adding an entry to the
batch so that the flush part doesn't have to get to the context.
This cleans it a bit, and paves the way to future updates like
dynamic vsids.
Olof Johansson [Tue, 20 Sep 2005 03:46:44 +0000 (13:46 +1000)]
[PATCH] ppc64: Updated Olof iommu updates 2/3
There are potential cases in the future where the IOMMU might be
mapping smaller pages than the regular MMU is using. Keep the
allocator working on MMU pagesizes, but the low-level mapping
functions need to map more than one TCE entry per page to deal with
this.
This patch enables SBus support for the cs4231 sound driver.
It is tested on an Ultra2. Capture and playback both work.
I experienced lags and crashes using certain threaded
players like ogg123 and mp3blaster, while the former is
lagging far more. This behavior may be specific to SMP
systems. It is reproducable using the dummy sound card
driver. Sox works flawlessly.
Setting up the calculation of ptr in snd_cs4231_playback_pointer
was a bit strange. I got it to work by not incrementing the
[pc]_periods_sent counter when starting DMA the first time
in cs4231_dma_trigger. Therefore this dummy thing.
[TCP]: Set default congestion control correctly for incoming connections.
Patch from Joel Sing to fix the default congestion control algorithm
for incoming connections. If a new congestion control handler is added
(via module), it should become the default for new
connections. Instead, the incoming connections use reno. The cause is
incorrect initialisation causes the tcp_init_congestion_control()
function to return after the initial if test fails.
Cleanup the printk's in fib_trie:
* Convert a couple of places in the dump code to BUG_ON
* Put log level's on each message
The version message really needed the message since it leaks out
on the pretty Fedora bootup.
[AF_PACKET]: Allow for > 8 byte hardware addresses.
The convention is that longer addresses will simply extend
the hardeware address byte arrays at the end of sockaddr_ll and
packet_mreq.
In making this change a small information leak was also closed.
The code only initializes the hardware address bytes that are
used, but all of struct sockaddr_ll was copied to userspace.
Now we just copy sockaddr_ll to the last byte of the hardware
address used.
For error checking larger structures than our internal
maximums continue to be allowed but an error is signaled if we can
not fit the hardware address into our internal structure.
[PATCH] nvidiafb: Fix absence of cursor in nvidiafb
A recent change in nvidiafb caused nvidiafb_cursor to always return -ENXIO
instead of using the soft_cursor. This will happen if the parameter "hwcur"
is not set, which happens to be the default.
Roland Dreier [Tue, 20 Sep 2005 17:52:04 +0000 (10:52 -0700)]
[PATCH] IPoIB: Don't flush workqueue from within workqueue
ipoib_mcast_restart_task() is always called from within the
single-threaded IPoIB workqueue, so flushing the workqueue from within
the function can lead to a recursion overflow. But since we're
running in a single-threaded workqueue, we're already synchronized
against other items in the workqueue, so just get rid of the flush in
ipoib_mcast_restart_task().
Russell King [Tue, 20 Sep 2005 16:52:13 +0000 (17:52 +0100)]
[ARM] Prevent deadlock in page fault handler
As per x86, we may deadlock while trying to get the mmap semaphore.
Implement the same fix, which allows (eg) recursive faults to cause
an oops instead of deadlocking.
Ben Dooks [Tue, 20 Sep 2005 15:45:20 +0000 (16:45 +0100)]
[ARM] 2927/1: .arch.info - postfix section with .init for `make buildcheck`
Patch from Ben Dooks
The `make buildcheck` is erroneously reporting that the .arch.info
list is referencing items in the .init section as it is not itself
postfixed with .init
Ben Dooks [Tue, 20 Sep 2005 15:35:03 +0000 (16:35 +0100)]
[ARM] 2926/1: .proc.info - postfix section with .init for `make buildcheck`
Patch from Ben Dooks
The `make buildcheck` is erroneously reporting that the .proc.info
list is referencing items in the .init section as it is not itself
postfixed with .init
Ben Dooks [Tue, 20 Sep 2005 15:25:12 +0000 (16:25 +0100)]
[ARM] 2925/3: earlyparam - postfix section with .init for `make buildcheck`
Patch from Ben Dooks
The `make buildcheck` is erroneously reporting that the earlyparam
list is referencing items in the .init section as it is not itself
postfixed with .init
Also, as per rmk's suggestion, rename the __early_param to
.early_param to bring it into line with everything else
David S. Miller [Tue, 20 Sep 2005 03:11:57 +0000 (20:11 -0700)]
[SPARC64]: Move DCACHE_ALIASING_POSSIBLE define to asm/page.h
This showed that arch/sparc64/kernel/ptrace.c was not getting
the define properly, and thus the code protected by this ifdef
was never actually compiled before. So fix that too.
Ed L. Cashin [Tue, 20 Sep 2005 02:57:36 +0000 (19:57 -0700)]
[BYTEORDER]: Document alignment and byteorder macros
This patch comments the fact that although passing le64_to_cpup et
al. is within the intended use of the byteorder macros, using
get_unaligned is the recommended way to go.
Because we use byte loads/stores to cons up the value
in and out of registers, we can't expect the ASI endianness
setting to take care of this for us. So do it by hand.
This case is triggered by drivers/block/aoe/aoecmd.c in the
ataid_complete() function where it goes:
/* word 100: number lba48 sectors */
ssize = le64_to_cpup((__le64 *) &id[100<<1]);
This &id[100<<1] address is 4 byte, rather than 8 byte aligned,
thus triggering the unaligned exception.
projects / linux.git / log