Henrique de Moraes Holschuh wrote:
On Thu, 19 Sep 2013, Nick Lewycky wrote:
Henrique de Moraes Holschuh wrote:
I am trying to reproduce this issue with AMD, so as to try to track down
which kernels are affected (and if possibly, track down WHAT fixed the bug
as that isn't clear right now).
Did the issue happen on a *first install* of amd64-microcode, or did it
happen only when you tried to upgrade the package from an older version ?
It worked on first install and only broke when I tried to upgrade.
Can you check your kernel logs and tell me whether that first install caused
a processor microcode update?
Sorry, no dice.
Or failing that, if you could please purge the package, reboot (so that
you'll be running with BIOS microcode) and install it again (which would do
a "first microcode update"), just to make sure first installs are safe?
This worked! (For clarification, I purged, cold rebooted and installed
2.20131007.1+really20130710.1.)
It may be that the problem happens when you upgrade the processor microcode
twice in a row, i.e. first installs would never cause it to happen, and the
first update done by the initramfs would also be always safe.
Not applicable, I don't have an initramfs.
Also, if you upgraded the kernel to solve the issue, which version you're
using (and did it fix the issue) ?
No, I haven't gotten around to upgrading my kernel yet. I'm
convinced by your explanation that the kernel is buggy and hope to
get to that soon (this weekend?).
Well, if you do, 3.10.x is supposed to be safe, but I'd really appreciate if
you could test it like this:
1. install the old microcode package[1], with the microcode module loaded,
and check /var/log/kern.log to verify whether a microcode update really
happened. You might need to trigger themicrocode update manually with
(for kernel 3.10):
echo -n 1> /sys/devices/system/cpu/microcode/reload
(for kernels before 3.6):
echo -n 1> /sys/devices/system/cpu/cpu0/microcode/reload.
microcode module was already loaded at boot, installing
amd64-microcode_2.20120910-1_amd64.deb has this to say in the kern.log:
Sep 29 18:00:31 calvin kernel: ld-linux-x32.so[5878] general protection
ip:f771ee8d sp:ffe62ee8 error:0 in ld-2.17.so[f7708000+21000]
Sep 29 18:00:33 calvin kernel: ld-linux-x32.so[6574] general protection
ip:f7766e8d sp:ffdf2538 error:0 in ld-2.17.so[f7750000+21000]
Sep 29 18:00:37 calvin kernel: ld-linux-x32.so[7212] general protection
ip:f77cee8d sp:ffbefa68 error:0 in ld-2.17.so[f77b8000+21000]
. Running the echo command didn't add anything to kern.log.
2. upgrade the microcode package to the new one in testing/unstable, which
should update the microcode *again* (check in /var/log/kern.log, and if
necessary please reissue the "reload" trigger above) to be sure.
apt-get install amd64-microcode picks version
_2.20131007.1+really20130710.1_amd64 whose installation *succeeds* and
adds this to kern.log:
Sep 29 18:01:46 calvin kernel: microcode: CPU0: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU1: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU2: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU3: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU4: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU5: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU6: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU7: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU8: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU9: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU10: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU11: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU12: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU13: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU14: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU15: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU16: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU17: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU18: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU19: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU20: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU21: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU22: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU23: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU24: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU25: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU26: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU27: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU28: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU29: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU30: new patch_level=0x06000822
Sep 29 18:01:46 calvin kernel: microcode: CPU31: new patch_level=0x06000822
Sep 29 18:01:48 calvin kernel: ld-linux-x32.so[8002] general protection
ip:f773ee8d sp:ffce0638 error:0 in ld-2.17.so[f7728000+21000]
Sep 29 18:01:49 calvin kernel: ld-linux-x32.so[8698] general protection
ip:f7756e8d sp:ffb17328 error:0 in ld-2.17.so[f7740000+21000]
Sep 29 18:01:53 calvin kernel: ld-linux-x32.so[9336] general protection
ip:f77b6e8d sp:ff8c37c8 error:0 in ld-2.17.so[f77a0000+21000]
Let me try dpkg --remove 'ing it:
Sep 29 18:03:02 calvin kernel: ld-linux-x32.so[10004] general protection
ip:f777ee8d sp:ffe38498 error:0 in ld-2.17.so[f7768000+21000]
Sep 29 18:03:03 calvin kernel: ld-linux-x32.so[10687] general protection
ip:f774ee8d sp:ffcc44d8 error:0 in ld-2.17.so[f7738000+21000]
Sep 29 18:03:07 calvin kernel: ld-linux-x32.so[11325] general protection
ip:f776ee8d sp:ffa7a218 error:0 in ld-2.17.so[f7758000+21000]
and then --purge 'ing it:
Sep 29 18:03:28 calvin kernel: ld-linux-x32.so[11970] general protection
ip:f7786e8d sp:ff8a09a8 error:0 in ld-2.17.so[f7770000+21000]
Sep 29 18:03:29 calvin kernel: ld-linux-x32.so[12653] general protection
ip:f76f6e8d sp:ffc4fe38 error:0 in ld-2.17.so[f76e0000+21000]
Sep 29 18:03:32 calvin kernel: ld-linux-x32.so[13291] general protection
ip:f77bee8d sp:ffaaf018 error:0 in ld-2.17.so[f77a8000+21000]
and then reinstalling the new version with apt:
Sep 29 18:04:14 calvin kernel: ld-linux-x32.so[14074] general protection
ip:f77dee8d sp:ffcd4498 error:0 in ld-2.17.so[f77c8000+21000]
Sep 29 18:04:15 calvin kernel: ld-linux-x32.so[14770] general protection
ip:f772ee8d sp:ffe92888 error:0 in ld-2.17.so[f7718000+21000]
Sep 29 18:04:18 calvin kernel: ld-linux-x32.so[15408] general protection
ip:f77b6e8d sp:ffd49508 error:0 in ld-2.17.so[f77a0000+21000]
and the installation completed successfully.
Nick
--
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]
