Marcelo Tosatti: Linux 2.4.26-pre3


Full Changelog


Here goes the third pre release of 2.4.26.

It contains a bunch of NFS client fixes, IA64/32-bit PPC
updates, sisfb update, amongst others.


Summary of changes from v2.4.26-pre2 to v2.4.26-pre3


  • sb16 sample size fix


  • Updates to the bootloader code due to changes from immap to
  • Add the RPC/EP 8260 board to the configuration for testing the
    new cpm2 updates.


  • Fix LVM snapshot oversized sector calculation


  • mremap NULL pointer dereference fix


  • Fix vmalloc() spinlocking issues introduced by the “error
    handling fixes”

Adrian Bunk:

  • agpgart_be.c: remove duplicate PCI_DEVICEIDSI_651

Andrew Morton:

  • [CRYPTO]: arc4.c compile fix for older gcc’s

Arun Sharma:

  • ia64: Fix and optimize sys32_rt_sigtimedwait()

Bjorn Helgaas:

  • ia64: (acpi_hp_csr_space): Export only if CONFIG_ACPI
  • ia64: Tidy up MCA printk’s
  • ia64: (desc_abi): Check for Linux ABI # (3) instead of SysV4
    ABI # (0)
  • ia64: unwind: Add some UNW_DEBUG stuff and remove KDB bits to
    follow 2.6
  • ia64: Use __builtin_trap() in BUG() when available
  • ia64: Remove obsolete sigcontext comment

David Mosberger:

  • ia64: Drop copyright notices on header files which are either
    entirely trivial
  • ia64: Back-port from libunwind: fix off-by-one error in
  • ia64: Fix bug in ia64_getscratchnat_bits()/ia64_putscratchnat_bits()

David S. Miller:

  • [TIGON3]: tg3_phy_copper_begin() tweaks
  • [TIGON3]: Allow MAC address changing even when iface is up
  • [TIGON3]: Always force PHY reset after major hw config
  • [TIGON3]: Update driver version and reldate

David Stevens:

  • [IPV4]: Add sysctl for per-socket limit on number of mcast src
  • [IPV4/IPV6]: Add sysctl limits for mcast src filters

Grant Grundler:

  • [TIGON3]: Consolidate MMIO write flushing using tg3_f()

Keith Owens:

  • ia64: Avoid deadlock when using printk() for MCA and INIT
  • ia64: Delete all MCA/INIT/etc record printing code, moved to
    salinfo_decode in user space.
  • ia64: Mark MCA variables and functions static where
  • ia64: Delete dead variables and functions from mca.c
  • ia64: Reorder mca.c to remove the need for forward declarations
    and to consolidate related code.
  • ia64: Synchronize mca.c with 2.6. White space, comment, #ifdef,
  • ia64: MCA, salinfo: calculate irq_safe once and pass it
  • ia64: Correct “did we recover from MCA” test and move up a
  • ia64: Periodically check for outstanding MCA or INIT
  • ia64: remove include/asm-ia64/offset.hs in “make mrproper”
  • ia64: Delete redundant ia64_mca_check_errors
  • ia64: update unwind with 2.6 fixes
  • ia64: add OEM data decode for SGI MCA handler
  • ia64: copy SAL records so we don’t drop them when events occur

Kumar Gala:

  • [PPC32] imported in CPM 8260_io drivers from linuxppc_2_4_devel
  • [PPC32] Added support ADS 8272 board
  • [PPC32] Renamed 8260 CPM handling to CPM2. This is to allow
    reuse of CPM2 code between PQ2 and PQ3 devices. 8xx is considered
  • [PPC32] Renamed 8260 CPM handling to CPM2. This allows reuse of
    CPM2 code between PQ2 and PQ3 devices. 8xx is considered CPM1

Marcelo Tosatti:

  • Ogawa Hirofumi: fix FAT over NFSv2
  • Change LAN media MAINTAINERS entry to Orphan
  • Changed EXTRAVERSION to -pre3

Patrick McHardy:

  • [PKT_SCHED]: Fix ipv6 ECN marking in RED scheduler

Thomas Winischhofer:

  • sisfb update

Trond Myklebust:

  • A patch to fix an Oops in the locking code
  • Slight optimization to the NFS writes
  • A patch by Greg Banks that increases the supported NLM cookie
    size. This is needed in order to work correctly with Apple and
    FreeBSD clients.
  • A patch by Patrice Dumas that ensures that the server index
    blocks uniquely by using the client address in addition to the
    value of the NLM cookie field.
  • A patch by Greg Banks to help fix the “VFS: Busy inodes after
    unmount.” problem that occurs if autofs expires the mountpoint
    while an NFS sillydelete is still pending.
  • I have a feeling the second race case of your test is that you
    are interrupting the fcntl(F_SETLK) call while it is on the wire.
    If you do that, then the server may record the lock as taken, but
    the client will never receive the reply, and so will not know that
    it must clean up locks…

Wensong Zhang:

  • [IPVS]: Code tidy up

Get the Free Newsletter!

Subscribe to Developer Insider for top news, trends, & analysis