kernel_optimize_test/net/rds
Linus Torvalds 799c10559d De-pessimize rds_page_copy_user
Don't try to "optimize" rds_page_copy_user() by using kmap_atomic() and
the unsafe atomic user mode accessor functions.  It's actually slower
than the straightforward code on any reasonable modern CPU.

Back when the code was written (although probably not by the time it was
actually merged, though), 32-bit x86 may have been the dominant
architecture.  And there kmap_atomic() can be a lot faster than kmap()
(unless you have very good locality, in which case the virtual address
caching by kmap() can overcome all the downsides).

But these days, x86-64 may not be more populous, but it's getting there
(and if you care about performance, it's definitely already there -
you'd have upgraded your CPU's already in the last few years).  And on
x86-64, the non-kmap_atomic() version is faster, simply because the code
is simpler and doesn't have the "re-try page fault" case.

People with old hardware are not likely to care about RDS anyway, and
the optimization for the 32-bit case is simply buggy, since it doesn't
verify the user addresses properly.

Reported-by: Dan Rosenberg <drosenberg@vsecurity.com>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-10-15 11:09:28 -07:00
..
af_rds.c net: sk_sleep() helper 2010-04-20 16:37:13 -07:00
bind.c
cong.c Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6 2010-04-11 14:53:53 -07:00
connection.c include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h 2010-03-30 22:02:32 +09:00
ib_cm.c net/rds: Add missing mutex_unlock 2010-05-29 00:18:48 -07:00
ib_rdma.c Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6 2010-04-11 14:53:53 -07:00
ib_recv.c Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6 2010-04-11 14:53:53 -07:00
ib_ring.c
ib_send.c RDS: Properly unmap when getting a remote access error 2010-03-16 21:17:00 -07:00
ib_stats.c
ib_sysctl.c sysctl: Drop & in front of every proc_handler. 2009-11-18 08:37:40 -08:00
ib.c include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h 2010-03-30 22:02:32 +09:00
ib.h RDS/IB+IW: Move recv processing to a tasklet 2009-10-30 15:06:39 -07:00
info.c include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h 2010-03-30 22:02:32 +09:00
info.h
iw_cm.c net/rds: Add missing mutex_unlock 2010-05-29 00:18:48 -07:00
iw_rdma.c include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h 2010-03-30 22:02:32 +09:00
iw_recv.c Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6 2010-04-11 14:53:53 -07:00
iw_ring.c
iw_send.c RDS: Do not BUG() on error returned from ib_post_send 2010-03-16 21:16:53 -07:00
iw_stats.c
iw_sysctl.c sysctl: Drop & in front of every proc_handler. 2009-11-18 08:37:40 -08:00
iw.c include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h 2010-03-30 22:02:32 +09:00
iw.h RDS/IB+IW: Move recv processing to a tasklet 2009-10-30 15:06:39 -07:00
Kconfig
loop.c Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6 2010-04-11 14:53:53 -07:00
loop.h
Makefile
message.c include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h 2010-03-30 22:02:32 +09:00
page.c De-pessimize rds_page_copy_user 2010-10-15 11:09:28 -07:00
rdma_transport.c Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6 2010-04-27 12:49:13 -07:00
rdma_transport.h
rdma.c Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6 2010-04-11 14:53:53 -07:00
rdma.h RDS: Add GET_MR_FOR_DEST sockopt 2009-10-30 15:06:37 -07:00
rds.h net: sk_sleep() helper 2010-04-20 16:37:13 -07:00
recv.c rds: fix a leak of kernel memory 2010-08-18 23:40:03 -07:00
send.c net: sk_sleep() helper 2010-04-20 16:37:13 -07:00
stats.c
sysctl.c sysctl: Drop & in front of every proc_handler. 2009-11-18 08:37:40 -08:00
tcp_connect.c net: fix a lockdep splat 2010-09-24 22:26:10 -07:00
tcp_listen.c net: fix a lockdep splat 2010-09-24 22:26:10 -07:00
tcp_recv.c net: fix a lockdep splat 2010-09-24 22:26:10 -07:00
tcp_send.c net: fix a lockdep splat 2010-09-24 22:26:10 -07:00
tcp_stats.c
tcp.c include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h 2010-03-30 22:02:32 +09:00
tcp.h
threads.c RDS: Enable per-cpu workqueue threads 2010-03-16 21:17:02 -07:00
transport.c