提交 · 70674f95c0a2ea694d5c39f4e514f538a09be36f · openeuler / Kernel

29 3月, 2006 37 次提交

[PATCH] Optimize select/poll by putting small data sets on the stack · 70674f95

由 Andi Kleen 提交于 3月 28, 2006

Optimize select and poll by a using stack space for small fd sets

This brings back an old optimization from Linux 2.0. Using the stack is
faster than kmalloc. On a Intel P4 system it speeds up a select of a
single pty fd by about 13% (~4000 cycles -> ~3500)

It also saves memory because a daemon hanging in select or poll will
usually save one or two less pages. This can add up - e.g. if you have 10
daemons blocking in poll/select you save 40KB of memory.

I did a patch for this long ago, but it was never applied. This version is
a reimplementation of the old patch that tries to be less intrusive. I
only did the minimal changes needed for the stack allocation.

The cut off point before external memory is allocated is currently at
832bytes. The system calls always allocate this much memory on the stack.

These 832 bytes are divided into 256 bytes frontend data (for the select
bitmaps of the pollfds) and the rest of the space for the wait queues used
by the low level drivers. There are some extreme cases where this won't
work out for select and it falls back to allocating memory too early -
especially with very sparse large select bitmaps - but the majority of
processes who only have a small number of file descriptors should be ok.
[TBD: 832/256 might not be the best split for select or poll]

I suspect more optimizations might be possible, but they would be more
complicated. One way would be to cache the select/poll context over
multiple system calls because typically the input values should be similar.
Problem is when to flush the file descriptors out though.
Signed-off-by: NAndi Kleen <ak@suse.de>
Cc: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

70674f95

[PATCH] ide_generic_all_on() warning fix · b02389e9

由 Andrew Morton 提交于 3月 28, 2006

drivers/ide/pci/generic.c:45: warning: `ide_generic_all_on' defined but not used

Cc: Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

b02389e9

[PATCH] Small fixes backported to old IDE SiS driver · d266ab88

由 Alan Cox 提交于 3月 28, 2006

Some quick backport bits from the libata PATA work to fix things found in
the sis driver.  The piix driver needs some fixes too but those are way to
large and need someone working on old IDE with time to do them.

This patch fixes the case where random bits get loaded into SIS timing
registers according to the description of the correct behaviour from
Vojtech Pavlik.  It also adds the SiS5517 ATA16 chipset which is not
currently supported by the driver.  Thanks to Conrad Harriss for loaning me
the machine with the 5517 chipset.
Signed-off-by: NAlan Cox <alan@redhat.com>
Acked-by: NBartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

d266ab88

[PATCH] ide: AMD756 no host side cable detection · 9edc91df

由 Rene Herman 提交于 3月 28, 2006

>From http://marc.theaimsgroup.com/?l=linux-kernel&m=110304128900342&w=2

AMD756 doesn't support host side cable detection. Do disk side only and
don't advice obsolete options.
Acked-by: NBartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

9edc91df

[PATCH] autofs4: proper prototype for autofs4_dentry_release() · f5b95ff0

由 Adrian Bunk 提交于 3月 28, 2006

Add a proper prototype for autofs4_dentry_release() to autofs_i.h.
Signed-off-by: NAdrian Bunk <bunk@stusta.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

f5b95ff0

[PATCH] drivers/block/acsi_slm.c: size_t can't be < 0 · fa8f399a

由 Alexey Dobriyan 提交于 3月 28, 2006

A size_t can't be < 0.

(akpm: and rw_verify_area() already did that check)
Signed-off-by: NAdrian Bunk <bunk@stusta.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

fa8f399a

[PATCH] remove relayfs_fs.h · e5123609

由 Andrew Morton 提交于 3月 28, 2006

This is obsolete.

Cc: Tom Zanussi <zanussi@us.ibm.com>
Cc: Jens Axboe <axboe@suse.de>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

e5123609

[PATCH] Philip Gladstone has moved · 3ca1da4a

由 Philip Gladstone 提交于 3月 28, 2006

I noticed that my email address is four jobs ago.
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

3ca1da4a

[PATCH] fs/fat/: proper prototypes for two functions · a28af471

由 Adrian Bunk 提交于 3月 28, 2006

Add proper prototypes for fat_cache_init() and fat_cache_destroy() in
msdos_fs.h.
Signed-off-by: NAdrian Bunk <bunk@stusta.de>
Acked-by: NOGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

a28af471

[PATCH] drivers/block/paride/pd.c: fix an off-by-one error · 3b71797e

由 Adrian Bunk 提交于 3月 28, 2006

The Coverity checker found this off-by-one error.
Signed-off-by: NAdrian Bunk <bunk@stusta.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

3b71797e

[PATCH] Change dash2underscore() return value to char · b9e20a92

由 Eric Sesterhenn 提交于 3月 28, 2006

Since dash2underscore() just operates and returns chars, I guess its safe
to change the return value to a char.  With my .config, this reduces its
size by 5 bytes.

   text    data     bss     dec     hex filename
   4155     152       0    4307    10d3 params.o.orig
   4150     152       0    4302    10ce params.o
Signed-off-by: NEric Sesterhenn <snakebyte@gmx.de>
Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: NAdrian Bunk <bunk@stusta.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

b9e20a92

[PATCH] mqueue comment typo fix · 7b7a317c

由 Serge E. Hallyn 提交于 3月 28, 2006

(akpm: I don't do comment typos patches.  This one snuck through by accident)
Signed-off-by: NSerge Hallyn <serue@us.ibm.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

7b7a317c

[PATCH] alpha: make poll flags the same as other architectures · 7a1e524a

由 Andrew Morton 提交于 3月 28, 2006

Renumber the recently-added POLLREMOVE and POLLRDHUP to line up with the other
architectures.

Cc: Davide Libenzi <davidel@xmailserver.org>
Cc: Ulrich Drepper <drepper@redhat.com>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Richard Henderson <rth@twiddle.net>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

7a1e524a

[PATCH] Add oprofile_add_ext_sample · 27357716

由 Brian Rogan 提交于 3月 28, 2006

On ppc64 we look at a profiling register to work out the sample address and
if it was in userspace or kernel.

The backtrace interface oprofile_add_sample does not allow this.  Create
oprofile_add_ext_sample and make oprofile_add_sample use it too.
Signed-off-by: NAnton Blanchard <anton@samba.org>
Cc: Philippe Elie <phil.el@wanadoo.fr>
Cc: John Levon <levon@movementarian.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

27357716

[PATCH] symversion warning fix · f83ca9fe

由 Andrew Morton 提交于 3月 28, 2006

gcc-4.2:

kernel/module.c: In function '__find_symbol':
kernel/module.c:158: warning: the address of '__start___kcrctab', will always evaluate as 'true'
kernel/module.c:165: warning: the address of '__start___kcrctab_gpl', will always evaluate as 'true'
kernel/module.c:182: warning: the address of '__start___kcrctab_gpl_future', will always evaluate as 'true'
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

f83ca9fe

[PATCH] capi: register_chrdev() fix · 6d9eac34

由 Andrew Morton 提交于 3月 28, 2006

If the user specified `major=0' (odd thing to do), capi.c will use dynamic
allocation.  We need to pick up that major for subsequent unregister_chrdev().
Acked-by: NKarsten Keil <kkeil@suse.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

6d9eac34

[PATCH] paride-pt: register_chrdev fix · 829d5f68

由 Andrew Morton 提交于 3月 28, 2006

If the user specified `major=0' (odd thing to do), pt.c will use dynamic
allocation. We need to pick up that major for subsequent unregister_chrdev().
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

829d5f68

[PATCH] paride: register_chrdev fix · 8637980b

由 Andrew Morton 提交于 3月 28, 2006

If the user specified `major=0' (odd thing to do), pg.c will use dynamic
allocation. We need to pick up that major for subsequent unregister_chrdev().
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

8637980b

[PATCH] let BLK_DEV_RAM_COUNT depend on BLK_DEV_RAM · a687fb18

由 Adrian Bunk 提交于 3月 28, 2006

It's purely cosmetic, but with the patch there's no longer a
BLK_DEV_RAM_COUNT setting in the .config if BLK_DEV_RAM=n.
Signed-off-by: NAdrian Bunk <bunk@stusta.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

a687fb18

[PATCH] synclink_gt: remove uneeded async code · be4676e6

由 Paul Fulghum 提交于 3月 28, 2006

Remove code in async receive handling that serves no purpose with new tty
receive buffering.  Previously this code tried to free up receive buffer
space, but now does nothing useful while making expensive calls.
Signed-off-by: NPaul Fulghum <paulkf@microgate.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

be4676e6

[PATCH] synclink_gt add gpio feature · 0080b7aa

由 Paul Fulghum 提交于 3月 28, 2006

Add driver support for general purpose I/O feature of the Synclink GT
adapters.
Signed-off-by: NPaul Fulghum <paulkf@micrgate.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

0080b7aa

[PATCH] synclink: remove dead code · 86a34147

由 Paul Fulghum 提交于 3月 28, 2006

Remove dead code from synclink driver.  This was used previously when the
write method had a from_user flag, which has been removed.
Signed-off-by: NPaul Fulghum <paulkf@microgate.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

86a34147

[PATCH] Decrapify asm-generic/local.h · f5f5370d

由 Kyle McMartin 提交于 3月 28, 2006

Now that Christoph Lameter's atomic_long_t support is merged in mainline,
might as well convert asm-generic/local.h to use it, so the same code can
be used for both sizes of 32 and 64-bit unsigned longs.

akpm sayeth:

Q:

  Is there any particular reason why these routines weren't simply
  implemented with local_save/restore_flags, if they are only meant to
  guarantee atomicity to the local cpu?  I'm sure on most platforms this
  would be more efficient than using an atomic...

A:

  The whole _point_ of local_t is to avoid local_irq_disable().  It's
  designed to exploit the fact that many CPUs can do incs and decs in a way
  which is atomic wrt local interrupts, but not atomic wrt SMP.

But this patch makes sense, because asm-generic/local.h is just a fallback
implementation for architectures which either cannot perform these
local-irq-atomic operations, or its maintainers haven't yet got around to
implementing them.

We need more work done on local_t in the 2.6.17 timeframe - they're defined as
unsigned long, but some architectures implement them as signed long.
Signed-off-by: NKyle McMartin <kyle@parisc-linux.org>
Cc: Benjamin LaHaise <bcrl@kvack.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

f5f5370d

[PATCH] RTC: Remove some duplicate BCD definitions · 4f3a36a7