URL:
http://www.us.kernel.org/pub/linux/kernel/people/andrea/kernels/v2.4/2.4.21rc2aa1.gz
Diff between 2.4.21pre5aa2 and 2.4.21rc2aa1:
Only in 2.4.21rc2aa1: 00_01_cciss-1
Only in 2.4.21rc2aa1: 00_02_cciss-1
Only in 2.4.21rc2aa1: 00_03_cciss-1
CCISS driver updates found on l-k.
Only in 2.4.21pre5aa2: 00_conntrack-hash-1
Only in 2.4.21pre5aa2: 00_i810-unlock_page-fastcall-1
Only in 2.4.21pre5aa2: 30_00_fix_symlink-1
Only in 2.4.21pre5aa2: 30_01_fix_softirq-1
Only in 2.4.21pre5aa2: 30_17-fix_read-1
Only in 2.4.21pre5aa2: 72_swapl-1
Merged in mainline.
Only in 2.4.21pre5aa2: 00_cpu-affinity-syscall-rml-3
Only in 2.4.21rc2aa1: 00_cpu-affinity-syscall-rml-4
Only in 2.4.21pre5aa2: 00_drop-inetpeer-cache-3.gz
Only in 2.4.21rc2aa1: 00_drop-inetpeer-cache-4.gz
Only in 2.4.21pre5aa2: 00_extraversion-21
Only in 2.4.21rc2aa1: 00_extraversion-22
Only in 2.4.21pre5aa2: 00_sched-O1-aa-2.4.19rc3-10.gz
Only in 2.4.21rc2aa1: 00_sched-O1-aa-2.4.19rc3-11.gz
Only in 2.4.21pre5aa2: 00_smp-timers-not-deadlocking-2
Only in 2.4.21rc2aa1: 00_smp-timers-not-deadlocking-3
Only in 2.4.21pre5aa2: 05_vm_23_per-cpu-pages-1
Only in 2.4.21rc2aa1: 05_vm_23_per-cpu-pages-2
Only in 2.4.21pre5aa2: 10_sched-o1-hyperthreading-3
Only in 2.4.21rc2aa1: 10_sched-o1-hyperthreading-4
Only in 2.4.21pre5aa2: 20_pte-highmem-29.gz
Only in 2.4.21rc2aa1: 20_pte-highmem-30.gz
Only in 2.4.21pre5aa2: 30_p3-p4-optimize-1
Only in 2.4.21rc2aa1: 30_p3-p4-optimize-2
Only in 2.4.21pre5aa2: 60_tux-syscall-4
Only in 2.4.21rc2aa1: 60_tux-syscall-5
Only in 2.4.21rc2aa1: 82_x86_64-suse-11
Only in 2.4.21pre5aa2: 82_x86_64-suse-9
Only in 2.4.21pre5aa2: 84_x86-64-arch-2
Only in 2.4.21rc2aa1: 84_x86-64-arch-3
Only in 2.4.21pre5aa2: 9910_shm-largepage-11.gz
Only in 2.4.21rc2aa1: 9910_shm-largepage-12.gz
Only in 2.4.21pre5aa2: 9950_futex-3.gz
Only in 2.4.21rc2aa1: 9950_futex-4.gz
Only in 2.4.21pre5aa2: 9995_frlock-gettimeofday-4
Only in 2.4.21rc2aa1: 9995_frlock-gettimeofday-5
Only in 2.4.21pre5aa2: 9999_gcc-3.3-3
Only in 2.4.21rc2aa1: 9999_gcc-3.3-6
Rediffed.
Only in 2.4.21rc2aa1: 00_fix-end_kio_request-race-1
Must wakeup with end_buffer_io_kiobuf as last thing to be sure the
wakeup will happen on a still allocated kiobuf. Fixed by
Chris Mason.
Only in 2.4.21pre5aa2: 00_netconsole-2.4.10-C2-2.gz
Only in 2.4.21rc2aa1: 00_netconsole-2.4.10-C2-3.gz
netconsole fixes from Andi Kleen.
Only in 2.4.21rc2aa1: 00_radeon-3
Only in 2.4.21pre5aa2: 00_radeon-Mobility9000-2
More updates from Margit.
Only in 2.4.21rc2aa1: 00_remove_inode_page-prune_icache-smp-race-1
Fix mm corrupting SMP race between remove_inode_page and prune_icache.
Found by Chris Mason.
Only in 2.4.21rc2aa1: 00_tcp-spurious-dupack-winup-streamers-1
Avoid suprious duplicate acks for very minor window updates,
that generates the double outgoing traffic with streaming
services that tends to fill the whole receive window to buffer.
Only in 2.4.21rc2aa1: 00_tcp-tw-death-2
Fix from Olof Johansson for tcp_tw_death_row corruption.
Only in 2.4.21rc2aa1: 00_usbnet-zaurus-c700-1
Allow the new zaurus to connect with usbnet (from Frank Bennett).
Only in 2.4.21pre5aa2: 00_vma-file-merge-2
Only in 2.4.21rc2aa1: 00_vma-file-merge-3
Fix two bugs, first a fd leak found by Stephen C. Tweedie,
forgive me for the stupid email I sent as second follow up ;).
The second bug is a vma merging issue with device driver supplied
mappings for example with a ->close method, as worse we should
call its close method, it's simpler and safer to forbid merging
with those special things. vma file merging is really only worthwhile
with the normal file mappings.
Only in 2.4.21rc2aa1: 05_vm_24_accessed-ipi-only-smp-1
Avoid the tlb flushes for the invalid bit in the pagetables
if in SMP, they certainly ain't worthwhile.
Only in 2.4.21pre5aa2: 07_qlogicfc-2.gz
Only in 2.4.21rc2aa1: 07_qlogicfc-4.gz
Only in 2.4.21pre5aa2: 08_qlogicfc-template-aa-5
Only in 2.4.21rc2aa1: 08_qlogicfc-template-aa-6
Only in 2.4.21pre5aa2: 09_qlogic-link-2
Only in 2.4.21rc2aa1: 09_qlogic-link-3
qlogic driver update (this also does 64bit dma).
Only in 2.4.21rc2aa1: 21_pte-highmem-mremap-smp-scale-1
Backported from 2.5 the kmap_atomic functionalty in mremap
that is more scalable in smp.
Only in 2.4.21rc2aa1: 30_17-mmap-fix-1
Truncate mappings before updating the attributes on the
server, otherwise page faults re-dirtying pages can happen
between the ->setattr() and the vmtruncate(). From Trond Myklebust.
Only in 2.4.21pre5aa2: 9900_aio-18.gz
Only in 2.4.21rc2aa1: 9900_aio-20.gz
Various minor fixes.
Only in 2.4.21rc2aa1: 9985_blk-atomic-7
Only in 2.4.21pre5aa2: 9985_blk-atomic-aa6
Stop troubles using BH_Atomic to advertize if the elv_seq is
significant or not (from Jens Axboe).
Add atomic guarantee to asynchronous I/O too (before this patch
the atomic guarantee was provided only with regular (synchronous)
rawio, now it will be corruption-proof also with rawio+async).
Thanks to the design of blk-atomic extending the support to asyncio
is been a two liner.
Only in 2.4.21rc2aa1: 9999_ia64-__clear_bit-1
Use the right (efficient/inchoerent) __clear_bit.
Only in 2.4.21rc2aa1: 9999_mmap-cache-1
Add mmap cache to take faster some common case (band-aid to hide
the O(N) complexity of mmap). Tries to get right 64bit archs in
32bit emulation with /proc/<pid>/mapped_base functional.
Andrea
Andrea Arcangeli wrote:
> Only in 2.4.21rc2aa1: 00_remove_inode_page-prune_icache-smp-race-1
>
> Fix mm corrupting SMP race between remove_inode_page and prune_icache.
> Found by Chris Mason.
Any chance this will get into mainline before 2.4.21?
Regards,
Carl-Daniel
On Thu, May 15, 2003 at 12:39:40AM +0200, Carl-Daniel Hailfinger wrote:
> Andrea Arcangeli wrote:
>
> > Only in 2.4.21rc2aa1: 00_remove_inode_page-prune_icache-smp-race-1
> >
> > Fix mm corrupting SMP race between remove_inode_page and prune_icache.
> > Found by Chris Mason.
>
> Any chance this will get into mainline before 2.4.21?
it's obviously safe so I think yes:
--- x/mm/filemap.c.~1~ 2003-04-24 16:37:50.000000000 +0200
+++ x/mm/filemap.c 2003-04-24 17:05:10.000000000 +0200
@@ -100,9 +100,10 @@ static inline void remove_page_from_inod
if (mapping->a_ops->removepage)
mapping->a_ops->removepage(page);
- mapping->nrpages--;
list_del(&page->list);
page->mapping = NULL;
+ wmb();
+ mapping->nrpages--;
}
static inline void remove_page_from_hash_queue(struct page * page)
Marcelo please apply, thanks!
Andrea