On Fri, 2011-06-03 at 14:24 -0700, Matt Carlson wrote: > On Fri, Jun 03, 2011 at 01:56:42PM -0700, Matt Carlson wrote: > > On Thu, Jun 02, 2011 at 08:13:55PM -0700, Ben Hutchings wrote: > > > On Sat, 2011-05-21 at 22:06 +0100, Ben Hutchings wrote: > > > > The Debian kernel team regularly backports driver updates to the Linux > > > > kernel in stable releases to add support for new hardware. In the > > > > current stable release, the Linux kernel is based on longterm series > > > > 2.6.32.y. > > > > > > > > We generally prefer to cherry-pick bug fixes and new hardware support, > > > > but there are so many interrelated changes to tg3 since 2.6.32 that > > > > this seems to be impossible. So I've prepared a backport of tg3 from > > > > Linux 2.6.38, which is available in the git branch: > > > > > > > > git://git.kernel.org/pub/scm/linux/kernel/git/bwh/debian-linux.git netdev-test > > > > > > > > The kernel configuration files we use are at > > > > <http://kernel.alioth.debian.org/config/2.6.32-33/> (note, this site > > > > will be unavailable for the next few hours). > > > > > > > > We would appreciate any help Broadcom can provide in testing this, and > > > > any advice on changes that should be added or reverted. > > > > > > Full source and binary packages containing these and other backported > > > drivers can now be found at: > > > > > > http://people.debian.org/~benh/packages/ > > > > > > I have not yet received any testing feedback, and without that we will > > > have to defer any updates to Debian 6.0.3 (about another 3 months away). > > > > > > Ben. > > > > Hi Ben. I'm going over the changes right now, and it looks like there > > are a lot of critical fixes that are missing. Standby and I'll try to > > generate a list of patches to include. > > Oh. Never mind. I didn't realize 57765 and later devices are not > enabled. My understanding is that the BCM57765 was supported by tg3 in Linux 2.6.38 and so will be supported by this backported version. > Here are a list of patches you might want to apply that affect > the rest of the devices: > > > commit fe234f0e5cbb880792d2d1ac0743cf8c07e9dde3 > tg3: Fix tg3_poll_controller() passing wrong pointer to tg3_interrupt() We already got this through stable update 2.6.32.11. If you're looking at the source package, you'll need to run 'debian/rules source' to generate fully patched source in debian/build/source. You'll probably find it easier to view the changes in the git branch though. > commit 28b041139e344ecd0f144d6205b004ae354cfa1e > net: preserve ifreq parameter when calling generic phy_mii_ioctl(). This is an API/ABI change so we don't want it. Did I accidentally include a change that depends on it? (I would expect at least a compiler warning in that case.) > commit 2ffcc981d823a0518c627ca22d51ef72d0b7ca9a > tg3: Set tx bug flags for more devices > commit 34eea5ac214353ccd93ef7dd8dbd10aed87f5f46 > tg3: Only allow phy ioctls while netif_running > commit 01c3a3920f9f78866420b2004602944fca45083a > tg3: Fix NVRAM selftest > commit aba49f2421d5287692aee961ab4ce2981fdf4939 > tg3: Disable MAC loopback test for CPMU devices > commit 49692ca1e686970bac5726c3fd925427bb3ae89d > tg3: Fix loopback tests OK. Have you had a chance to test the backported driver on an appropriate range of supported devices? Ben. -- Ben Hutchings Once a job is fouled up, anything done to improve it makes it worse.
Attachment:
signature.asc
Description: This is a digitally signed message part