.hwstart.c 8139too: fix HW initial flow http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=3f9738f73ad08ef770df64f145007bd27ac2fa16 .mac_addr.c 8139too: allow to set mac address on running device http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=bda6a15a0d283d531b865fb7c596bb3ff258e87e .cb_ref.c 8139too: use err.h macros http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=85920d43bd5ae8a96f434704438b57c0b8c31198 "Compile tested only" .convert_nops.c 8139: convert to net_device_ops http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=48dfcde46ef23cfad1e0c0f13d4eae15c7f46efc .netstats.c 8139too: use netstats in net_device structure http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=e1eac92e4cea32b5254d0b835de16d11e589502c .trans_start.c 8139too: don't set trans_start in start_xmit routine - collateral evolution http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=1ae5dc342ac78d7a42965fd1f323815f6f5ef2c1#patch10 Possible reversions: Revert "veth: prevent oops caused by netdev destructor" http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=11687a1099583273a8a98ec42af62b5bb5a69e45 http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=ae0e8e82205c903978a79ebf5e31c670b61fa5b4 [SCSI] cnic: fix error: implicit declaration of function `__symbol_get' http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=bc3bf8fd330ce981ce632a1a4a283eee46838f32 http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=7cbdca23c8a4e6d007b62c9136ba0e5f86e069d0 Other reversions that don't work so well: Revert "skge: fix ram buffer size calculation" http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=279e1dab949d33737557babfe9f74e0b74fbe39a This one has some kind of write ordering issue. The problem only occurs during data transfers of 50MB/s or more. The patch incorrectly changed the order in which some writes to the device occur. sky2: revert config space change http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=a40ccc6868943e74ec12f26a266ce1d0373b2b32 Original patch removed a bunch of device write operations to make PCI config space read-only. This had some unintentended effect. sky2: revert shutdown changes http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=3d1454dd93e84ad1394b6b1646f13795e9f6928e Patch removed some device read/write operations that "broke shutdown" on some systems. Revert "sky2: remove check for PCI wakeup setting from BIOS" http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=be63a21c9573fbf88106ff0f030da5974551257b Some BIOS's break if Wake On Lan is enabled, and the machine can't boot. Better to have some user's have to call ethtool to enable WOL than to break a single user's boot. sky2: revert to access PCI config via device space http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=b32f40c4853cdbe3d722a959fb0dd1ea048b50d0 "Performance" optimization which also "avoids issues with MMCONFIG etc." Revert "[netdrvr e100] experiment with doing RX in a similar manner to eepro100" http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=ca93ca428b8e09973f19e2725bf19cb3f1836034 "Unfortunately the hardware doesn't work exactly like this and therefore this patch actually breaks e100. Reverting the change brings it back to the previously known good state for 2.6.22." r8169: revert 7da97ec96a0934319c7fbedd3d38baf533e20640 (bis repetita) http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=50d84c2dc00e48ff9ba018ed0dd23276cf79e566 Screws up the hardware (requires cold-restart) r8169: revert part of 6dccd16b7c2703e8bbf8bca62b5cf248332afbe2 http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=c946b3047205d7e107be16885bbb42ab9f10350a Performance regression for some chips r8169: revert bogus BMCR reset http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=9e0db8ef4a8c8fd6f3a506259975d7f8db962421 Screws up hardware Revert "r8169: mac address change support" http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=209ad53bc19667a128d9c51beba873a5c62bff6e Original patch adds some hw writes, but these break some normal use cases. Revert "[BNX2]: Seems to not need net/tcp.h" http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=de081fa517fed81b0369f2e90ca87c30182879c8 Driver did not compile after original patch was applied Revert "net: ifb error path loop fix" http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=bcdddfb66cc998252d34758ce4109cedc0d24a5c Original code was convoluted and the patch author was confused. Executing this code might have revealed the problem. This driver is not for real hardware (ifb.c) [PATCH] sky2: revert IRQ dance on suspend/resume http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=4cbf2aa35e1c189db234190fefc6c83b139ef963 Original patch was fine, a bit messy, but they didn't want to deal with this problem this way. [PATCH] sky2: revert pci express extensions http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=7bd656d12119708b37414bf909ab2995473da818 Original patch writes different stuff to HW. [PATCH] forcedeth: fix multi irq issues http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=84b3932bf0fd8cdc8c75a5be77e1dded1e6479c6 Original patch does "request_irq" twice, accidentally, and then only frees it once. When the driver is unloaded, one IRQ line remains active, and this causes crashes. [PATCH] via-rhine: revert "change mdelay to msleep and remove from ISR path" http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=38bb6b288bf4fb954a3793e57c7339774c842a54 msleep vs. mdelay. This one could theoretically be checkable via symbolic execution. The problem is the call stack netdev_ioctl -> generic_mii_ioctl -> mdio_read -> rhine_disable_linkmon -> msleep(1); (vs. mdelay(1)). Since netdev_ioctl grabs a spinlock, the msleep will fail, but mdelay is OK. [TG3]: Revert "Speed up SRAM access" http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=758a613936223699f99081fc75eff7cf67b208f6 Hardware misbehaves w/ original patch e100: revert CPU cycle saver microcode, it causes severe problems http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=875521ddccfa90d519cf31dfc8aa472f7f6325bb Unclear as to the problem Notes: - revert "r8169: read MAC address from EEPROM on init (2nd attempt)" * Interesting because it's hardware-specific. * Bug is not fail-stop, so symexec wouldn't pick it up.