David Howells [Thu, 25 Oct 2007 14:27:40 +0000 (15:27 +0100)]
Fix pointer mismatches in proc_sysctl.c
Fix pointer mismatches in proc_sysctl.c. The proc_handler() method returns a
size_t through an arg pointer, but is given a pointer to a ssize_t to return
into.
Linus Torvalds [Thu, 18 Oct 2007 03:18:32 +0000 (23:18 -0400)]
Fix /proc/acpi/alarm BCD alarm encodings
This fixes some totally illogical and wrong code that converts things to
and from BCD mode essentially randomly, does math on values in BCD mode
etc etc. Introduce a few helper functions to make it a bit more obvious
what is going on, and make sure that we always do all the arithmetic
(and anythign else, for that matter) in binary, not BCD.
Tested by Mark Lord, who found the problem originally, and also pushed
the patch back and reminded me about it.
Sam Ravnborg [Thu, 25 Oct 2007 18:31:19 +0000 (20:31 +0200)]
x86: move defconfig files for i386 and x86_64 to x86
With some small changes to kconfig makefile we can now
locate the defconfig files for i386 and x86_64 in
the configs/ subdirectory under x86.
make ARCH=i386 defconfig and make defconfig
works as expected also after this change.
But arch maintainers shall now update a defconfig file in
the configs/ directory.
Jiri Kosina [Fri, 19 Oct 2007 22:05:19 +0000 (00:05 +0200)]
USB: usbserial - fix potential deadlock between write() and IRQ
USB: usbserial - fix potential deadlock between write() and IRQ
usb_serial_generic_write() doesn't disable interrupts when taking port->lock,
and could therefore deadlock with usb_serial_generic_read_bulk_callback()
being called from interrupt, taking the same lock. Fix it.
Oliver Neukum [Tue, 23 Oct 2007 10:26:41 +0000 (12:26 +0200)]
USB: fix read vs. disconnect race in cytherm driver
the disconnect method of this driver set intfdata to NULL before
removing attribute files. The attributes' read methods will happily
follow the NULL pointer. Here's the correct ordering.
Alan Stern [Fri, 19 Oct 2007 13:51:58 +0000 (09:51 -0400)]
USB: fix interface sysfs file-creation bug
This patch (as1005) fixes a rather subtle problem. When
usb_set_configuration() registers the interfaces and their files in
sysfs, it doesn't expect those files to exist already. But when an
interface is registered, its driver may call usb_set_interface() and
thereby cause the sysfs files to be created. The result is an error
when usb_set_configuration() goes on to create those same files again.
The (not-so-great) solution is to have usb_set_configuration() remove
any existing files before creating them.
Ingo Molnar [Mon, 15 Oct 2007 17:43:21 +0000 (19:43 +0200)]
USB: fix ssb_ohci_probe() build bug
fix ssb_ohci_probe() build bug:
drivers/built-in.o: In function `ssb_ohci_probe':
ohci-hcd.c:(.text+0xbff39): undefined reference to `ssb_device_enable'
ohci-hcd.c:(.text+0xbff6f): undefined reference to `ssb_admatch_base'
ohci-hcd.c:(.text+0xbff8b): undefined reference to `ssb_admatch_size'
ohci-hcd.c:(.text+0xbffe5): undefined reference to `ssb_device_disable'
[...]
the reason was that this Kconfig combination was allowed:
Alan Cox [Mon, 15 Oct 2007 14:08:11 +0000 (15:08 +0100)]
USB: isd200: sort out USB/IDE dependancy mess
The ISD200 driver imports a single trivial routine from the IDE layer and
in doing so creates a mess of dependancies that drag in the entire old
IDE layer. Even more sad - it does this for a routine which is usually
(little endian) a null function!
- Copy the function into ISD200
- Rename it so it doesn't clash with the ide header prototype
- Remove all the depend constraints
Alan Stern [Tue, 16 Oct 2007 15:55:30 +0000 (11:55 -0400)]
USB: fix scheduling of Iso URBs in uhci-hcd
This patch (as1003) changes uhci-hcd to treat the URB_ISO_ASAP flag
the same as other host controller drivers, namely, to schedule an Iso
URB for the first available time slot that hasn't already expired.
URBs in which the flag isn't set will be scheduled for the first slot
following the last URB, even if it has expired.
Andrew Morton [Thu, 18 Oct 2007 08:24:25 +0000 (01:24 -0700)]
USB: ftd_sio cleanups and updates for new termios work checkpatch fixes
WARNING: line over 80 characters
#23: FILE: drivers/usb/serial/ftdi_sio.c:297:
+ speed_t force_baud; /* if non-zero, force the baud rate to this value */
ERROR: use tabs not spaces
#31: FILE: drivers/usb/serial/ftdi_sio.c:881:
+^I$
ERROR: use tabs not spaces
#39: FILE: drivers/usb/serial/ftdi_sio.c:890:
+^I$
WARNING: line over 80 characters
#111: FILE: drivers/usb/serial/ftdi_sio.c:1956:
+ tty_encode_baud_rate(port->tty, priv->force_baud, priv->force_baud);
Your patch has style problems, please review. If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.
Alan Cox [Thu, 18 Oct 2007 08:24:19 +0000 (01:24 -0700)]
USB: digi_acceleport: fix termios and also readability a bit
- Expand some x&y to x & y so I could read it when checking
- Clear CMSPAR bit in the termios (as the driver does not support it)
- Encode the speed using the new tty_encode_baud_rate facility
Alan Cox [Thu, 18 Oct 2007 08:24:18 +0000 (01:24 -0700)]
USB: ch341: fix termios handling
The ch341 currently doesn't support most of the hardware setting. So to keep
the termios data right we propogate the old termios hardware values back then
encode the speed.
Alan Cox [Thu, 18 Oct 2007 08:24:16 +0000 (01:24 -0700)]
serial: stop passing NULL to functions that expect data
Earlier patches have removed the checking for old v new differences from the
USB drivers so we can now pass in a valid blank old termios so that we don't
to fill the drivers with magic hacks for console support
Hugh Dickins [Thu, 25 Oct 2007 07:55:05 +0000 (09:55 +0200)]
fix sg_phys to use dma_addr_t
x86_32 CONFIG_HIGHMEM64G with 5GB RAM hung when booting, after issuing
some "request_module: runaway loop modprobe binfmt-0000" messages in
trying to exec /sbin/init.
The binprm buf doesn't see the right ".ELF" header because sg_phys()
is providing the wrong physical addresses for high pages: a 32-bit
unsigned long is too small in this case, we need to use dma_addr_t.
Eugene Konev [Wed, 24 Oct 2007 02:42:02 +0000 (10:42 +0800)]
cpmac: convert to napi_struct interface
Convert cpmac to new napi_struct API introduced by bea3348eef27e6044b6161fd04c3152215f96411 [NET]: Make NAPI polling independent
of struct net_device objects.
Only disable rx interrupts if napi actually has been scheduled.
Eugene Konev [Wed, 24 Oct 2007 02:42:01 +0000 (10:42 +0800)]
cpmac: use print_mac() instead of MAC_FMT
Switch to using DECLARE_MAC_BUF/print_mac() added by commit 0795af5729b18218767fab27c44b1384f72dc9ad [NET]: Introduce and use print_mac()
and DECLARE_MAC_BUF().
> Hi,
>
> I booted up 2.6.24-rc1 this morning [Real early over a brew ;-)] and
> was having a problems with multiple ~5 second hangs on SATA/drive init
> (Something to do with "EH" something-or-other and resets but I'll
> email in separately about it later unless its fixed by the time I get
> the chance).
>
> Anyway, I went to fire up netconsole to get a decent log dump and hit
> across the following nasty. Netconsole works fine in 2.6.23.1 with a
> similar config and the same kernel parameters.
>
> A shot of the screen is the only method I could come up with to
> capture the log, I hope that is OK, it is pretty readable.
>
>
> The nasty:
> http://andotnet.nfshost.com/linux/2.6.24-rc1-netconsole-nullderef.jpg
[NET]: Make NAPI polling independent of struct net_device objects.
added np->dev to drivers/net/natsemi.c's struct netdev_private, but
forgot to initialize that new field upon driver init. The result was
a predictable NULL dereference oops the first time the hardware
generated an interrupt.
FUJITA Tomonori [Thu, 25 Oct 2007 07:13:32 +0000 (09:13 +0200)]
x86: pci-gart fix
map_sg could copy the last sg element to another position (if merging
some elements). It breaks sg chaining. This copies only
dma_address/length instead of the whole sg element.
Tejun Heo [Thu, 25 Oct 2007 05:59:16 +0000 (14:59 +0900)]
ahci: ahci: implement workaround for ASUS P5W-DH Deluxe ahci_broken_hardreset(), take #2
P5W-DH Deluxe has ICH9 which doesn't have PMP support but SIMG 4726
hardwired to the second port of AHCI controller at PCI device 1f.2.
The 4726 doesn't work as PMP but as a storage processor which can do
hardware RAID on downstream ports.
When no device is attached to the downstream port of the 4726, pseudo
ATA device for configuration appears. Unfortunately, ATA emulation on
the device is very lousy and causes long hang during boot.
This patch implements workaround for the board. If the mainboard is
P5W-DH Deluxe (matched using DMI), only hardreset is used on the
second port of AHCI controller @ 1f.2 and the hardreset doesn't depend
on receiving the first FIS and just proceed to IDENTIFY.
Tejun Heo [Wed, 24 Oct 2007 06:21:26 +0000 (15:21 +0900)]
libata: cosmetic clean up in ata_eh_reset()
Local variable @action usage in ata_eh_reset() is a bit confusing.
It's used only to cache ehc->i.action to test reset masks after
clearing it; however, due to the generic name "action", it's easy to
misinterpret the local variable as containing the selected reset
method later. Also, the reason for caching the original value is easy
to miss.
This patch renames @action to @tmp_action and make it buffer newly
selected value instead to improve readability.
Rusty Russell [Thu, 25 Oct 2007 04:09:25 +0000 (14:09 +1000)]
lguest: example launcher header cleanup.
Now the kernel headers are clean for userspace export, we don't need
to typedef kernel types before including them. We also don't need
pci_ids.h (that was from an earlier virtio draft).
Peter Williams [Wed, 24 Oct 2007 16:23:51 +0000 (18:23 +0200)]
sched: reduce balance-tasks overhead
At the moment, balance_tasks() provides low level functionality for both
move_tasks() and move_one_task() (indirectly) via the load_balance()
function (in the sched_class interface) which also provides dual
functionality. This dual functionality complicates the interfaces and
internal mechanisms and makes the run time overhead of operations that
are called with two run queue locks held.
This patch addresses this issue and reduces the overhead of these
operations.
profile=sleep only works if CONFIG_SCHEDSTATS is set. This patch notes
the limitation in Documentation/kernel-parameters.txt and prints a
warning at boot-time if profile=sleep is used without CONFIG_SCHEDSTAT.
Satyam Sharma [Wed, 24 Oct 2007 16:23:50 +0000 (18:23 +0200)]
sched: use show_regs() to improve __schedule_bug() output
A full register dump along with stack backtrace would make the
"scheduling while atomic" message more helpful. Use show_regs() instead
of dump_stack() for this. We already know we're atomic in here (that is
why this function was called) so show_regs()'s atomicity expectations
are guaranteed.
Also, modify the output of the "BUG: scheduling while atomic:" header a
bit to keep task->comm and task->pid together and preempt_count() after
them.
Milton Miller [Wed, 24 Oct 2007 16:23:48 +0000 (18:23 +0200)]
sched: fix sched_domain sysctl registration again
commit 029190c515f15f512ac85de8fc686d4dbd0ae731 (cpuset
sched_load_balance flag) was not tested SCHED_DEBUG enabled as
committed as it dereferences NULL when used and it reordered
the sysctl registration to cause it to never show any domains
or their tunables.
Fixes:
1) restore arch_init_sched_domains ordering
we can't walk the domains before we build them
presently we register cpus with empty directories (no domain
directories or files).
2) make unregister_sched_domain_sysctl do nothing when already unregistered
detach_destroy_domains is now called one set of cpus at a time
unregister_syctl dereferences NULL if called with a null.
While the the function would always dereference null if called
twice, in the previous code it was always called once and then
was followed a register. So only the hidden bug of the
sysctl_root_table not being allocated followed by an attempt to
free it would have shown the error.
3) always call unregister and register in partition_sched_domains
The code is "smart" about unregistering only needed domains.
Since we aren't guaranteed any calls to unregister, always
unregister. Without calling register on the way out we
will not have a table or any sysctl tree.
4) warn if register is called without unregistering
The previous table memory is lost, leaving pointers to the
later freed memory in sysctl and leaking the memory of the
tables.
Before this patch on a 2-core 4-thread box compiled for SMT and NUMA,
the domains appear empty (there are actually 3 levels per cpu). And as
soon as two domains a null pointer is dereferenced (unreliable in this
case is stack garbage):
bu19a:~# ls -R /proc/sys/kernel/sched_domain/
/proc/sys/kernel/sched_domain/:
cpu0 cpu1 cpu2 cpu3
Gerrit Renker [Wed, 24 Oct 2007 12:53:01 +0000 (10:53 -0200)]
[CCID2/3]: Initialisation assignments of 0 are redundant
Assigning initial values of `0' is redundant when loading a new CCID structure,
since in net/dccp/ccid.c the entire CCID structure is zeroed out prior to
initialisation in ccid_new():
Gerrit Renker [Wed, 24 Oct 2007 12:27:48 +0000 (10:27 -0200)]
[DCCP]: Convert Reset code into socket error number
This adds support for converting the 11 currently defined Reset codes into system
error numbers, which are stored in sk_err for further interpretation.
This makes the externally visible API behaviour similar to TCP, since a client
connecting to a non-existing port will experience ECONNREFUSED.
* Code 0, Unspecified, is interpreted as non-error (0);
* Code 1, Closed (normal termination), also maps into 0;
* Code 2, Aborted, maps into "Connection reset by peer" (ECONNRESET);
* Code 3, No Connection and
Code 7, Connection Refused, map into "Connection refused" (ECONNREFUSED);
* Code 4, Packet Error, maps into "No message of desired type" (ENOMSG);
* Code 5, Option Error, maps into "Illegal byte sequence" (EILSEQ);
* Code 6, Mandatory Error, maps into "Operation not supported on transport endpoint" (EOPNOTSUPP);
* Code 8, Bad Service Code, maps into "Invalid request code" (EBADRQC);
* Code 9, Too Busy, maps into "Too many users" (EUSERS);
* Code 10, Bad Init Cookie, maps into "Invalid request descriptor" (EBADR);
* Code 11, Aggression Penalty, maps into "Quota exceeded" (EDQUOT)
which makes sense in terms of using more than the `fair share' of bandwidth.
Gerrit Renker [Wed, 24 Oct 2007 12:18:06 +0000 (10:18 -0200)]
[DCCP]: One more exemption from full sequence number checks
This fixes the following problem: client connects to peer which has no DCCP
enabled or loaded; ICMP error messages ("Protocol Unavailable") can be seen
on the wire, but the application hangs. Reason: ICMP packets don't get through
to dccp_v4_err.
When reporting errors, a sequence number check is made for the DCCP packet
that had caused an ICMP error to arrive.
Such checks can not be made if the socket is in state LISTEN, RESPOND (which
in the implementation is the same as LISTEN), or REQUEST, since update_gsr()
has not been called in these states, hence the sequence window is 0..0.
This patch fixes the problem by adding the REQUEST state as another exemption
to the window check. The error reporting now works as expected on connecting.
Gerrit Renker [Wed, 24 Oct 2007 12:12:09 +0000 (10:12 -0200)]
[DCCP]: Retrieve packet sequence number for error reporting
This fixes a problem when analysing erroneous packets in dccp_v{4,6}_err:
* dccp_hdr_seq currently takes an skb
* however, the transport headers in the skb are shifted, due to the
preceding IPv4/v6 header.
Fixed for v4 and v6 by changing dccp_hdr_seq to take a struct dccp_hdr as
argument. Verified that the correct sequence number is now reported in the
error handler.
WANG Cong [Wed, 24 Oct 2007 11:07:11 +0000 (13:07 +0200)]
arch/um/drivers/ubd_kern.c: fix a building error
Fix this uml building error:
arch/um/drivers/ubd_kern.c: In function 'do_ubd_request':
arch/um/drivers/ubd_kern.c:1118: error: implicit declaration of function
'sg_page'
arch/um/drivers/ubd_kern.c:1118: warning: passing argument 6 of
'prepare_request' makes pointer from integer without a cast
make[1]: *** [arch/um/drivers/ubd_kern.o] Error 1
make: *** [arch/um/drivers] Error 2
Signed-off-by: WANG Cong <[email protected]>
Add sg_init_table() call as well.
projects / linux.git / log