Alan Cox: Linux 2.4.6-ac4

ftp://ftp.kernel.org/pub/linux/kernel/people/alan/2.4/. Intermediate diffs are available from http://www.bzimage.org.
Cox writes, “So far so good, but still treat this one with care. We now use the
official VIA workaround for the southbridge bugs. That should fix
the ES137x/SB PCI problems on VIA and some other stuff without
breaking the IDE corruption fix. The i810 security hole is now patched too.”

2.4.6-ac4
o Update VIA southbridge bug fix to VIA provided
workaround. (me)
| Except we apply it even when no sblive is present
o Fix up includes to use slab.h (Chris Wedgewood)
o Switch es1371 internal amp to a table (Frank Aune, me)
o Merge improve zone balancing (Marcelo Tosatti)
| Should fix a lot of high zone balancing problems
o Update to megaraid 1.17a (Atul Mukker)
o Add large file support to user mode Linux (Lennert Buytenhek)
o Add management console to user mode Linux (Lennert Buytenhek)
o IDE updates (Andre Hedrick)
o Updated SBNI drivers (Yaroslav Polyakov, Denis I.Timofeev)
o Fix the i810 ‘read any kernel memory’ bug (me)

2.4.6-ac3
o Save registers on pnpbios calls (me)
o Avoid re-entrancy on PnPBIOs calls (me)
o Catch oopsing in PnPBIOS and print better
diagnostics (ie rant about bios vendor)
(me)
o Fix CONFIG_ARM entry (me)
o Fix reference to major in mnd_pinnacle (me)
o Merge Linus 2.4.7pre6
o Fix dmi to return null string for string 0 (Matt Domsch)
o Don’t probe MCA serial space on non MCA x86 (me)
| Fixes the Debian combo MCA/PCI kernel problems
o Turbo pam isdn fix (Stelian Pop)
o Add ISAPnP PCMCIA adapter support (Andrey Panin)
o Fix NO_FSBR and ZERO_PACKET user urbs (Thomas Sailer)
o Fix types used in sk98 for save/restore flags (Jes Sorensen)
o Update Configure.help (Fumiaki OHATA)
o Fix scheduler/yield bug (Andrea Arcangeli)
o Ext2fs clean up of high/acl usage (Andreas Dilger)
o Fix chattr on ext2 directories (Andreas Dilger)
o Correct quota read/write check (Andreas Dilger)
o Fix i810_tco oops (Randy Dunlop)
o Update bootflag handler (Randy Dunlop)
o Fix errata/erratum comment errors (David Woodhouse)
o Further starfire fixes (Ion Badulescu)
o Fix FAT overflow due to signed i_pos (OGAWA Hirofumi)
o Fix reiserfs tree balance/filldir race (Chris Mason)
o USB check for vendor types in devio (Dan Streetman)
o Clean up the dynamic aic7xxx files in make clean(Tim Hockin)
o Fix pci ordering in headers (Tim Hockin)
o Add DMI detection for funny purple laptops (Stelian Pop)
o Bluetooth driver update (Greg Kroah-Hartmann)
o Fix pnpbios help width (Randy Dunlop)
o Edgeport driver update (Greg Kroah-Hartmann)
o Fix softirq pending breakage in UML (Jeff Dike)
o Add ID for CPIA2 cameras (Steve Miller)
o dmi printk fix (Robert Dunlop)
o USB storage wrong ID for SDDR-09 (Phil Stracchino)
o Add configure.help for VAIO LCD mode (Stelian Pop)
o Add larger readahead to the ataraid (Arjan van de Ven)
o Further ISAPnP idents for SB variants (Bill Nottingham)
o Minixfs updates (Al Viro)
o Frank Denis has moved email (Frank Denis)
o Support triple indirect shmem for S/390x (Christoph Rohland)
o shmem race fix (Christoph Rohland)
o Fix hole in binfmt_elf loader (Sebastian ,
Solar Designer)
o Merge farsync synchronous driver (Bob Dunlop)
o 8139too locking fix (Masaru Kawashima)
o Update do_wp_page documentation (Hugh Dickins)

o Fix zr36120 range check (me)
| This batch is all from the Stanford checker
o Fix stradis range check (me)
o Fix i2c maths overrun security check (me)
o Fix bttv range check (me)
o Fix zr36067 range check (me)
o Limit unique length to 1024 in drm (me)
o Bound nframes on cd-rom reading (me)
o Add missing range check on ov511 (me)
o Fix range checks in mga (me)
| I’ve no idea if I’ve broken the matrox driver in doing so
| but right now I don’t actually care. XFree need to fix it right
o Ditto for radeon (me)
o Ditto for r128 (me)
o Ditto for matrox pci (me)
o And generic drm_addbufs (me)
o Fix missing
o Fix wrong kfree order in ali-ircc (me)
o Fix use of freed object in CDCEther (me)
o Fix double free of urb in usbnet (me)
o Fix wrong free in usb_new_device (me)

* Note that the i810 drm bug reported by the checker looks exploitable.
It
needs someone from XFree86 with i810 docs to fix this. In the mean time
people should probably disable the i810 drm support if its an issue.