No point in adding yet another #ifdef for Loongson since all this mask is
being used for is converting an XKPHYS address into a physical address
anyway. So replace all definitions by one with the highest architectural
possible value.
Florian Fainelli [Tue, 22 May 2007 19:44:42 +0000 (21:44 +0200)]
[MIPS] Add generic GPIO to Au1x00
This patch adds support for the generic GPIO API to Au1x00 boards. It requires
the generic GPIO patch for MIPS boards by Yoichi Yuasa. Now there is a MIPS
target using it, can you queue these patchset for 2.6.22 ? Thank you very
much in advance.
Atsushi Nemoto [Tue, 29 May 2007 15:38:07 +0000 (00:38 +0900)]
[MIPS] Simplify missing-syscalls for N32 and O32
Use standard missing-syscalls with EXTRA_CFLAGS instead of duplicating
the command. And move the archprepare rule before the archclean rule.
Suggested by Franck Bui-Huu. Also add "echo" to show the target ABI.
Alan Cox [Tue, 10 Jul 2007 16:05:16 +0000 (17:05 +0100)]
IOC3: Switch to pci refcounting safe APIs
Convert the IOC3 driver to use ref counting pci interfaces so that we can
obsolete the (usually unsafe) pci_find_{slot/device} interfaces and avoid
future authors writing hotplug-unsafe device drivers.
David Graham [Tue, 10 Jul 2007 15:45:11 +0000 (08:45 -0700)]
e100: Fix Tyan motherboard e100 not receiving IPMI commands
The 82550 & 51 parts have an extended configuration block that
includes a bit "GMRC", required to enable the expected TCO behavior,
in config byte offset 22d. The config block sent by the failing driver
does include the extension area, but this bit is not initialised,
and the downlaod only specifies 0x16 bytes to be sent to the NIC
(thaht's bytes 00..21d). By initializing the GMRC bit, and extending
the download size for D102+ MACs, the problem is resolved.
We intend this patch to improve spidernet interrupt handling to be
more strict. We had following problem and this patch solves it.
-when CONFIG_DEBUG_SHIRQ=y, request_irq() calls handler().
-when spider_net_open() is called, it calls request_irq() which calls
spider_net_interrupt().
-if some specific interrupt bit is set at this timing, it calls
netif_rx_schedule() and spider_net_poll() is scheduled.
-spider_net_open() calls netif_poll_enable() which clears the bit
__LINK_STATE_RX_SCHED.
-when spider_net_poll() is called, it calls netif_rx_complete() which
causes BUG_ON() because __LINK_STATE_RX_SCHED is not set.
The 3cSOHO100-TX needs a mdio_sync() before mdio_read() to read the MII
transceiver registers properly. Adding EXTRA_PREAMBLE to drv_flags of the
3cSOHO100-TX will force this.
This problem exists already for years (I checked back to 2.6.8).
Setting duplex for the 3cSOHO100-TX was more or less a random process.
Till 2.6.15 it was more likely that the diver ends up in half
duplex mode, after the code change in 2.6.16 it was more likely
to end up in full duplex mode. I wonder why nobody noticed this earier.
This patch avoids generating another IRQ if more packets
arrive while in the NAPI poll routine. Before marking device as
finished, it rechecks that the status ring is empty.
This patch restores a couple of workarounds from 2.6.16:
* restart transmit moderation timer in case it expires during IRQ routine
* default to having 10 HZ watchdog timer.
At this point it more important not to hang than to worry about the
power cost.
Documentation/block/barrier.txt is not in sync with the actual code: - blk_queue_ordered() no longer has a gfp_mask parameter - blk_queue_ordered_locked() no longer exists - sd_prepare_flush() looks slightly different
With the cfq_queue hash removal, we inadvertently got rid of the
async queue sharing. This was not intentional, in fact CFQ purposely
shares the async queue per priority level to get good merging for
async writes.
So put some logic in cfq_get_queue() to track the shared queues.
On Tue, 2007-07-10 at 10:06 +0100, Christoph Hellwig wrote:
> > -#define GFS2_LARGE_FH_SIZE 10
> > -
> > -struct gfs2_fh_obj {
> > - struct gfs2_inum_host this;
> > - u32 imode;
> > -};
> > +#define GFS2_LARGE_FH_SIZE 8
>
> Because gfs2_decode_fh only accepts file handles with GFS2_LARGE_FH_SIZE
> or GFS2_LARGE_FH_SIZE you don't accept filehandles sent out by and older
> gfs version anymore. Stale filehandles because of a new kernel version
> are a big no-no, so please add back code to handle the old filehandles
> on the decode side.
>
This should fix that problem I think since its only relating to end of
the fh we can just ignore that field in order to accept the older
format.
Anton Vorontsov [Thu, 3 May 2007 20:32:17 +0000 (00:32 +0400)]
[BATTERY] pda_power platform driver
Common power driver for PDAs and phones with one or two external
power supplies (AC/USB) connected to main and backup batteries,
and optional builtin charger.
It's used to stop logic duplication through different embedded
devices. So, power supply *logic* is here. pda_power register
power supplies, and will take care about notifying batteries
about power changes through external power interface.
Currently, power consumption legal limits (including USB power
consumption) should be handled by platform code, inside set_charge
function.
sched-cfs-v2.6.22-git-v18.patch introduces CPU_IDLE in sched.h.
This conflict with the already existing define in
include/asm-s390/processor.h
Just rename the s390 defines, since they will go away as soon as
we support CONFIG_NO_HZ instead of our own CONFIG_NO_IDLE_HZ.
Ralph Wuerthner [Tue, 10 Jul 2007 09:24:19 +0000 (11:24 +0200)]
[S390] zcrypt: fix request timeout handling
Under very high load zcrypt requests may timeout while waiting on the
request queue. Modify zcrypt that timeouts are based on crypto adapter
responses. A timeout occurs only if a crypto adapter does not respond
within a given time frame to sumitted requests.
After the in-kernel system call has been remove the system call path
can be optimized. The problem state bit of the old psw is always set
between system_call and sysc_do_svc. SAVE_ALL_SVC uses this information
to avoid two instructions.
[S390] dasd: Avoid compile warnings on !CONFIG_DASD_PROFILE
drivers/s390/block/dasd_proc.c:33:
warning: 'dasd_get_user_string' defined but not used
drivers/s390/block/dasd_proc.c:172:
warning: 'dasd_statistics_array' defined but not used
Michael Holzheu [Tue, 10 Jul 2007 09:24:15 +0000 (11:24 +0200)]
[S390] Program check in diag 210 under 31 bit
If a virtual address is passed to the diag210 function under 31 bit, we get a
programming exception, since diag 210 only works with physical addresses. To
fix this, the content of the diag210 data structure is copied to a local
structure and the physical address of that structure is passed to diagnose 210.
The bogomips calculation triggered via reading from /proc/cpuinfo
can return incorrect values if the qrnnd assembly is called with a
pointer in %r2 with any of the upper 32 bits set.
Fix this by using 64 bit division / remainder operation provided by
gcc instead of calling the assembly.
Jens Axboe [Thu, 14 Jun 2007 11:10:48 +0000 (13:10 +0200)]
pipe: change the ->pin() operation to ->confirm()
The name 'pin' was badly chosen, it doesn't pin a pipe buffer
in the most commonly used sense in the kernel. So change the
name to 'confirm', after debating this issue with Hugh
Dickins a bit.
A good return from ->confirm() means that the buffer is really
there, and that the contents are good.
Carsten Otte [Fri, 15 Jun 2007 06:16:22 +0000 (08:16 +0200)]
xip sendfile removal
This patch removes xip_file_sendfile, the sendfile implementation for
xip without replacement. Those customers that use xip on s390 are not
using sendfile() as far as we know, and so far s390 is the only platform
this could potentially be used on so far.
Having sendfile is not a popular feature for execute in place file
systems, however we have a working implementation of splice_read() based
on fs/splice.c if anyone asks for it.
At this point in time, it does not seem preferable to merge
splice_read() for xip because it causes extra maintenence effort due to
code duplication and it requires struct page behind the xip memory
segment. We'd like to get rid of that in favor of supporting flash based
embedded platforms (Monta Vista work) soon.
Hugh Dickins [Mon, 4 Jun 2007 08:00:39 +0000 (10:00 +0200)]
shmem: convert to using splice instead of sendfile()
Remove shmem_file_sendfile and resurrect shmem_readpage, as used by tmpfs
to support loop and sendfile in 2.4 and 2.5. Now tmpfs can support splice,
loop and sendfile in the simplest way, using generic_file_splice_read and
generic_file_splice_write (with the aid of shmem_prepare_write).
We could make some efficiency tweaks later, if there's a real need;
but this is stable and works well as is.
Jens Axboe [Mon, 4 Jun 2007 07:59:47 +0000 (09:59 +0200)]
splice: divorce the splice structure/function definitions from the pipe header
We need to move even more stuff into the header so that folks can use
the splice_to_pipe() implementation instead of open-coding a lot of
pipe knowledge (see relay implementation), so move to our own header
file finally.
Jens Axboe [Thu, 14 Jun 2007 11:08:55 +0000 (13:08 +0200)]
vmsplice: add vmsplice-to-user support
A bit of a cheat, it actually just copies the data to userspace. But
this makes the interface nice and symmetric and enables people to build
on splice, with room for future improvement in performance.
Jens Axboe [Tue, 12 Jun 2007 19:17:17 +0000 (21:17 +0200)]
splice: abstract out actor data
For direct splicing (or private splicing), the output may not be a file.
So abstract out the handling into a specified actor function and put
the data in the splice_desc structure earlier, so we can build on top
of that.
This is the first step in better splice handling for drivers, and also
for implementing vmsplice _to_ user memory.
Boaz Harrosh [Wed, 20 Jun 2007 11:53:23 +0000 (13:53 +0200)]
[PATCH] fix request->cmd == INT cases
- I have unearthed very old bugs in stale drivers that still
used request->cmd as a READ|WRITE int
- This patch is maybe a proof that these drivers have not been
used for a long time. Should they be removed completely?
Drivers that currently do not work for sure:
drivers/acorn/block/fd1772.c | 2 +-
drivers/acorn/block/mfmhd.c | 8 ++++----
drivers/cdrom/aztcd.c | 2 +-
drivers/cdrom/cm206.c | 2 +-
drivers/cdrom/gscd.c | 2 +-
drivers/cdrom/mcdx.c | 2 +-
drivers/cdrom/optcd.c | 2 +-
drivers/cdrom/sjcd.c | 2 +-
Drivers with cosmetic fixes only:
b/drivers/block/amiflop.c
b/drivers/block/nbd.c
b/drivers/ide/legacy/hd.c
Tejun Heo [Tue, 19 Jun 2007 07:18:13 +0000 (09:18 +0200)]
[BLOCK] drop unnecessary bvec rewinding from flush_dry_bio_endio
Barrier bios are completed twice - once after the barrier write itself
is done and again after the whole sequence is complete.
flush_dry_bio_endio() is for the first completion. It doesn't really
complete the bio. It rewinds bvec and resets bio so that it can be
completed again when the whole barrier sequence is complete.
The bvec rewinding code has the following problems.
1. The rewinding code is wrong because filesystems may pass bvec with
non zero bv_offset.
2. The block layer doesn't guarantee anything about the state of
bvec array on request completion. bv_offset and len are updated
iff __end_that_request_first() completes the bvec partially.
Because of #2, #1 doesn't really matter (nobody cares whether bvec is
re-wound correctly or not) but then again by not doing unwinding at
all, we'll always give back the same bvec to the caller as full bvec
completion doesn't alter bvecs and the final completion is always full
completion.
projects / linux.git / log