lkml.org 
[lkml]   [2004]   [Jul]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
Subject2.4.23aa3
This includes only the pending important fixes (I started this on 31
May, and I finished it now that all needed fixes are already public).
Some user asked for an update like this. If you're still using 2.4.23aa2
you're recommended to upgrade to 2.6 or 2.4.23aa3 or the latest mainline
2.4.

Marcelo you may want to look into merging for
9999_z-get_user_pages_pte_pin-1, that's one relevant bugfix suitable for
mainline 2.4 too. It it rejects badly or you need more info on it please
don't hesitate to ask. Normally this bug goes unnoticed since most
people does rawio on shm where there are no COWs.
Thanks.

http://www.us.kernel.org/pub/linux/kernel/people/andrea/kernels/v2.4/2.4.23aa3.gz
http://www.us.kernel.org/pub/linux/kernel/people/andrea/kernels/v2.4/2.4.23aa3/

Changelog diff between 2.4.23aa2 and 2.4.23aa3:

Only in 2.4.23aa3: 00_acpi-bridge-swizzle-1
Only in 2.4.23aa3: 00_airo-proc-write-overflow-1
Only in 2.4.23aa3: 00_asus-acpi-sscanf-1
Only in 2.4.23aa3: 00_cgc-x86_64-transform-2
Only in 2.4.23aa3: 00_chown-DAC-check-missing-1
Only in 2.4.23aa3: 00_decnet-check-pointers-1
Only in 2.4.23aa3: 00_eql-check-dev-1
Only in 2.4.23aa3: 00_fix-e1000-infoleak-1
Only in 2.4.23aa3: 00_fix-ext3-infoleak-1
Only in 2.4.23aa3: 00_fix-fork-memleak-1
Only in 2.4.23aa3: 00_fix-hotplug-allocation-1
Only in 2.4.23aa3: 00_fix-jfs-infoleak-1
Only in 2.4.23aa3: 00_fix-soundblaster-dos-1
Only in 2.4.23aa3: 00_fwait-exceptions-1
Only in 2.4.23aa3: 00_gdth-pci-dma-2
Only in 2.4.23aa3: 00_highuid16-1
Only in 2.4.23aa3: 00_i810_dma-buf_count-1
Only in 2.4.23aa3: 00_iso9660-fix-1
Only in 2.4.23aa3: 00_lvm-vmalloc-1
Only in 2.4.23aa3: 00_mcast-msfilter-int-overflow-1
Only in 2.4.23aa3: 00_mpu401-user-pointer-1
Only in 2.4.23aa3: 00_nfs-autofs-umount-crash-1
Only in 2.4.23aa3: 00_nfsd-no-tcp-mixup-1
Only in 2.4.23aa3: 00_nfs-invalidate_inode_pages2-1
Only in 2.4.23aa3: 00_pss-user-pointer-1
Only in 2.4.23aa3: 00_r128-size-check-1
Only in 2.4.23aa2: 00_rawio-crash-1
Only in 2.4.23aa3: 00_rawio-crash-2
Only in 2.4.23aa3: 00_rtc-initialize-1
Only in 2.4.23aa3: 00_scsi-req-q-ptr-stall-1
Only in 2.4.23aa3: 00_sctp-fix-1
Only in 2.4.23aa3: 00_smbfs-uid32-1
Only in 2.4.23aa3: 00_sound-copy-to-user-1
Only in 2.4.23aa3: 00_sunrpc-svcsock-drop-1
Only in 2.4.23aa3: 00_symbios-fullspeed-1
Only in 2.4.23aa3: 00_tg3-sn2-1
Only in 2.4.23aa3: 00_usb-storage-ioerror_ohci-locking-1
Only in 2.4.23aa3: 00_x86_64-compute-csum-1
Only in 2.4.23aa3: 00_x86_64-erratum93-1
Only in 2.4.23aa3: 00_x86_64-iommu-fencepost-1
Only in 2.4.23aa3: 00_x86_64-iommu-fullflush-1
Only in 2.4.23aa3: 00_x86_64-ldt-limit-1
Only in 2.4.23aa3: 00_x86_64-mtrr-1
Only in 2.4.23aa3: 00_x86_64-node-limit-1
Only in 2.4.23aa3: 00_x86_64-ptrace32-1
Only in 2.4.23aa3: 00_x86_64-tiocgdev-fix-1
Only in 2.4.23aa3: 20_ext3-direct-io-fix-1
Only in 2.4.23aa3: 73_fix-xfs-infoleak-1
Only in 2.4.23aa3: 9999900_panic-overflow-1
Only in 2.4.23aa3: 9999900_qla-mknod-1
Only in 2.4.23aa3: 9999_fault-handler-infoleak-1
New fixes.

Only in 2.4.23aa3: 00_coredump-lfs-1

Allow >2G coredumps (mostly for 64bit archs).

Only in 2.4.23aa2: 00_extraversion-34
Only in 2.4.23aa3: 00_extraversion-35
Only in 2.4.23aa2: 90_proc-mapped-base-6
Only in 2.4.23aa3: 90_proc-mapped-base-7

Rediffed.

Only in 2.4.23aa2: 20_pte-highmem-32.gz
Only in 2.4.23aa3: 20_pte-highmem-33.gz

Fix unlikely to trigger vmalloc race.

Only in 2.4.23aa2: 9900_aio-25.gz
Only in 2.4.23aa3: 9900_aio-26.gz

Fix wtd semaphore race condition (I think it comes from some redhat
kernel).

Only in 2.4.23aa3: 9999900_pte-dirty-1
s390 will not mark the pte dirty during page fault (it works
with physical pages dirty flags not ptes dirty flags).

Only in 2.4.23aa3: 9999_z-get_user_pages_pte_pin-1
This is a major fix for mm corruption during COW with rawio.
I got convinced it was safe to leave pages not pinned into the
pte during the rawio, but it broke off with COWs that executes
copy-on-write while we write to the user page, and then the write
is lost since the copy happend in the middle of the I/O. In 2.6
I fixed it in a more elegant way that doesn't introduce any locking
thanks to the mapcount information (in 2.4 there's only the page->count)
and I used a bitflag to pin the pages, but this means the I/O is
serialized page against page, that's fine for 2.4.

Only in 2.4.23aa2: 9999_zzz-dynamic-hz-4.gz
Only in 2.4.23aa3: 9999_zzz-dynamic-hz-5.gz
Fix a silly (luckily minor) bug in the kernel-user hz conversion
routines (if you had any problem with the floppy or with timings with
some HZ value this will fix it).
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 14:04    [W:0.088 / U:0.508 seconds]
©2003-2014 Jasper Spaans. Advertise on this site