The file was in a subfolder amd-ucode-2012-01-17. With microcode-0.20120313-1.1.mga2 it should be found by microcode service on AMD K10 cpu's.
Please QA, test if you have an AMD K10. "dmesg | grep microcode" should help
Status: NEW => ASSIGNEDAssignee: bugsquad => qa-bugs
Created attachment 2408 [details] microcode nonfree/updates_testing commands/results (not a K10 cpu)
Created attachment 2409 [details] set -x in /etc/rc.d/init.d/microcode_ctl results
Workaround solution is to disable the minor check. # dif /etc/rc.d/init.d/microcode_ctl /etc/rc.d/init.d/microcode_ctl~ 42c42 < # if [ $minor -ge 29 ]; then --- > if [ $minor -ge 29 ]; then 49c49 < # fi --- > fi # systemctl start microcode_ctl.service # systemctl status microcode_ctl.service microcode_ctl.service - LSB: Update the Intel / AMD CPU microcode Loaded: loaded (/etc/rc.d/init.d/microcode_ctl) Active: active (exited) since Thu, 31 May 2012 06:16:02 -0500; 4min 18s ago CGroup: name=systemd:/system/microcode_ctl.service May 31 06:16:02 wb.home.test microcode_ctl[21348]: Loading AMD microcode update module May 31 06:16:02 wb.home.test microcode_ctl[21348]: [ OK ]
CC: (none) => junk_no_spam
(In reply to comment #2) > Created attachment 2408 [details] > microcode nonfree/updates_testing commands/results (not a K10 cpu) This is another bug : #4327 . This bug report is only to check that microcode is loaded, that's why there is no point in testing it without a K10.
No k10 here, this might need a mail to -discuss or -dev José Is there no way to test it without the hardware?
(In reply to comment #6) > No k10 here, this might need a mail to -discuss or -dev José Yes, feel free to do so : the patch is also applied in cauldron, so any cauldron user with a K10 has already tested it ;-) > > Is there no way to test it without the hardware? No.
Confirmed the directory fix i586 $ urpmf --media Release microcode | grep amd-ucode microcode:/lib/firmware/amd-ucode microcode:/lib/firmware/amd-ucode/amd-ucode-2012-01-17 microcode:/lib/firmware/amd-ucode/amd-ucode-2012-01-17/INSTALL microcode:/lib/firmware/amd-ucode/amd-ucode-2012-01-17/LICENSE microcode:/lib/firmware/amd-ucode/amd-ucode-2012-01-17/README microcode:/lib/firmware/amd-ucode/amd-ucode-2012-01-17/microcode_amd.bin microcode:/lib/firmware/amd-ucode/amd-ucode-2012-01-17/microcode_amd.bin.README microcode:/lib/firmware/amd-ucode/amd-ucode-2012-01-17/microcode_amd_fam15h.bin microcode:/lib/firmware/amd-ucode/amd-ucode-2012-01-17/microcode_amd_fam15h.bin.README microcode:/lib/firmware/amd-ucode/amd-ucode-2012-01-17/microcode_amd_solaris.bin microcode:/lib/firmware/amd-ucode/amd-ucode-2012-01-17/microcode_amd_solaris.bin.README $ urpmf --media Testing microcode | grep amd-ucode microcode:/lib/firmware/amd-ucode microcode:/lib/firmware/amd-ucode/INSTALL microcode:/lib/firmware/amd-ucode/LICENSE microcode:/lib/firmware/amd-ucode/README microcode:/lib/firmware/amd-ucode/microcode_amd.bin microcode:/lib/firmware/amd-ucode/microcode_amd.bin.README microcode:/lib/firmware/amd-ucode/microcode_amd_fam15h.bin microcode:/lib/firmware/amd-ucode/microcode_amd_fam15h.bin.README microcode:/lib/firmware/amd-ucode/microcode_amd_solaris.bin microcode:/lib/firmware/amd-ucode/microcode_amd_solaris.bin.README It will still need the correct hardware to properly test
dmesg | grep microcode gives exactly nothing. packages microcode_ctl is installed, but microcode itself isn't. i suppose i need to install microcode and regen dracut? then reboot and see if it works? the microcode_ctl service does fail according to systemctl though. To Confirm: this is for mga2 right? not for cauldron?
CC: (none) => alien
CC: (none) => hhielscher
(In reply to comment #9) > dmesg | grep microcode gives exactly nothing. I have see microcode output in dmesg, after I removed the test for kernel version. See comment 4. Go ahead and click up a terminal, run tail -f /var/log/messages then run systemctl start microcode_ctl.service and you will get about the same info you would have seen in dmesg. > packages microcode_ctl is installed, but microcode itself isn't. I did install microcode hoping it was the problem. Even after installing microcode and doing a systemctl start microcode.service, microcode_ctl was executed. > i suppose i need to install microcode and regen dracut? then reboot and see if > it works? I would guess not. > the microcode_ctl service does fail according to systemctl though. > > To Confirm: this is for mga2 right? not for cauldron? Yep, it was always failing for mga2 on two of my newer systems. My old testbed hardware did not complain when microcode_ctl ran.
after installing microcode, and restarting microcode_ctl (which failed), i got: [root@localhost ~]# dmesg | grep microcode [1205139.797425] microcode: CPU0: patch_level=0x010000bf [1205139.824905] microcode: failed to load file amd-ucode/microcode_amd.bin [1205139.824917] microcode: CPU1: patch_level=0x010000bf [1205139.825280] microcode: failed to load file amd-ucode/microcode_amd.bin [1205139.825301] microcode: CPU2: patch_level=0x010000bf [1205139.825685] microcode: failed to load file amd-ucode/microcode_amd.bin [1205139.825714] microcode: CPU3: patch_level=0x010000bf [1205139.825967] microcode: failed to load file amd-ucode/microcode_amd.bin [1205139.825972] microcode: CPU4: patch_level=0x010000bf [1205139.826261] microcode: failed to load file amd-ucode/microcode_amd.bin [1205139.826271] microcode: CPU5: patch_level=0x010000bf [1205139.826473] microcode: failed to load file amd-ucode/microcode_amd.bin [root@localhost ~]# urpmf amd-ucode/microcode_amd.bin [root@localhost ~]# urpmf /microcode_amd.bin microcode:/lib/firmware/amd-ucode/amd-ucode-2012-01-17/microcode_amd.bin microcode:/lib/firmware/amd-ucode/amd-ucode-2012-01-17/microcode_amd.bin.README so the paths are indeed different here too...
after installing the update candidate, and restarting the microcode_ctl service: [1205728.282937] microcode: CPU0: patch_level=0x010000bf [1205728.283640] microcode: CPU0: new patch_level=0x010000dc [1205728.283646] microcode: CPU1: patch_level=0x010000bf [1205728.284100] microcode: CPU1: new patch_level=0x010000dc [1205728.284106] microcode: CPU2: patch_level=0x010000bf [1205728.284407] microcode: CPU2: new patch_level=0x010000dc [1205728.284415] microcode: CPU3: patch_level=0x010000bf [1205728.284680] microcode: CPU3: new patch_level=0x010000dc [1205728.284691] microcode: CPU4: patch_level=0x010000bf [1205728.284962] microcode: CPU4: new patch_level=0x010000dc [1205728.284982] microcode: CPU5: patch_level=0x010000bf [1205728.285250] microcode: CPU5: new patch_level=0x010000dc so, i would say the update is valid for me on x86_64
I have a K10 AMD Phenom(tm) II X4 B50 Processor Confirmed that microcode does not load. Installed microcode-0.20120313-1.1.mga2.nonfree fron non-free/updates/testing Confirmed that microcode does load correctly # dmesg | grep microcode [288700.877796] microcode: CPU0: patch_level=0x010000b6 [288700.879834] microcode: CPU0: new patch_level=0x010000c8 [288700.879853] microcode: CPU1: patch_level=0x010000b6 [288700.880927] microcode: CPU1: new patch_level=0x010000c8 [288700.880955] microcode: CPU2: patch_level=0x010000b6 [288700.881961] microcode: CPU2: new patch_level=0x010000c8 [288700.881988] microcode: CPU3: patch_level=0x010000b6 [288700.882984] microcode: CPU3: new patch_level=0x010000c8 [288700.883312] microcode: Microcode Update Driver: v2.00 <tigran@aivazian.fsnet.co.uk>, Peter Oruba [288700.913809] microcode: Microcode Update Driver: v2.00 removed. However status returned by systemctl status microcode_ctl.service is incorrect # systemctl status microcode_ctl.service microcode_ctl.service - LSB: Update the Intel / AMD CPU microcode Loaded: loaded (/etc/rc.d/init.d/microcode_ctl) Active: failed (Result: exit-code) since Mon, 11 Jun 2012 14:11:57 +0100; 51s ago Process: 6472 ExecStart=/etc/rc.d/init.d/microcode_ctl start (code=exited, status=22) CGroup: name=systemd:/system/microcode_ctl.service Jun 11 14:11:57 Derek.localdomain microcode_ctl[6472]: [FAILED]
CC: (none) => derekjenn
Thanks Derek, are you using i586 or x86_64?
x86_64
i confirm that the exit code of the service is wrong (not a major issues, but someone should fix nonetheless), if there's ever dependencies on it.
boot.log reports that microcode_ctl has failed so if we do not fix it users are going to keep on asking why they have a failure in their log. The problem appears to be in /etc/rc.d/init.d/microcode_ctl It tests to see if the kernel minor release is greater than 29 before applying the microcode (obviously assuming the kernel is a 2.6 kernel) If I comment out the if statement it all works OK. case "$1" in start|reload|restart|force-reload) # Intel 686 and above, AMD family 16 and above vendor=`grep "^vendor_id" /proc/cpuinfo | head -n1 | awk -F ": " '{ print $2 }'` family=`grep "^cpu family" /proc/cpuinfo | head -n1 | awk -F ": " '{ print $2 }'` if [ "$vendor" = "GenuineIntel" ] && [ $family -ge 6 ]; then gprintf "Applying Intel CPU microcode update: " elif [ "$vendor" = "AuthenticAMD" ] && [ $family -ge 16 ]; then minor=`uname -r | cut -d . -f 3 | cut -d - -f 1` # if [ $minor -ge 29 ]; then gprintf "Loading AMD microcode update module\n" /sbin/modprobe microcode action "" /bin/true touch /var/lock/subsys/microcode_ctl /sbin/rmmod microcode exit 0 # fi else So do we release this upgrade to the microcode package and make a new bug for microcode_ctl, or do we fix them both at once?
Please read all the comments, or at least number 5 : bug #4327 already exists for the service return code, this bug is only about the firmware files path.
no offense, but even though it's mentioned, it seems there's more talk of intel microcode there... since mga3 will have systemd only, this only needs fixing in mga2, imho also the check should be more complete, just to avoid destroying CPU's. in any case, i've taken the liberty of making a commit to both cauldron and mga2 to fix this issue. if someone thinks this should get updated in mga2 or cauldron, they can bump release/subrel and submit and validate.
If the fix looks OK José do you want to submit it? I'm sure AL13N wouldn't mind retesting the new package and we could close both bugs with a working setup.
The fix is working on my Phenom II x6 1090T says updated all cores to newer firmware, but as above get the fail in my boot.log.
CC: (none) => lemonzest
(In reply to comment #20) > If the fix looks OK José do you want to submit it? > Ok so package microcode_ctl submitted : microcode_ctl-1.17-13.1.mga2 . When all will be validated, please push both microcode_ctl and microcode packages.
Tested on OK x86_64 AMD Phenom. Boot log is correct.
AL13N would you mind checking this fixes the problems you had previously please, this can then be validated. Thanks!
ping AL13N, this update is waiting for your test, if you can. Otherwise just tell and we'll find another way :)
Whiteboard: (none) => MGA2-64-OKCC: (none) => stormi
ah crap, forgot /o\ problem is that i almost never reboot. but i remember that manual changes did fix it for me...
(In reply to comment #26) > ah crap, forgot /o\ > > problem is that i almost never reboot. > > but i remember that manual changes did fix it for me... Does it mean that you can't test? If so, I think we'll validate the update as is.
i can't really test soon-ish (first, i'm not at home now) and secondly, i usually never reboot. so it'll be at least 1 or 2 months before i can test. thirdly, i have tried the same fix manually and that worked forme, so i assume this will go well. i agree with validation
Update validated. No linking required. SRPMS: - microcode_ctl-1.17-13.1.mga2 - microcode-0.20120313-1.1.mga2.nonfree Advisory coming in next comment.
CC: (none) => sysadmin-bugsKeywords: (none) => validated_update
un-validating until we get an advisory. AL13N or José, could you provide one? I'm not able to do it for this update.
Keywords: validated_update => (none)
Whiteboard: MGA2-64-OK => MGA2-64-OK MGA2-32-OK
Whiteboard: MGA2-64-OK MGA2-32-OK => MGA2-64-OK MGA2-32-OK feedback
Been using this since the update first went into testing on mga2 x86_64, results pretty much the same as above, now working fine :) (I have a Phenom II X6 1090T (K10.5))
possible advisory: Fix microcode startup script to handle 3.x kernels and add missing AMD CPU firmware.
Update validated. No linking required. SRPMS: - microcode_ctl-1.17-13.1.mga2 - microcode-0.20120313-1.1.mga2.nonfree --- Advisory --- The microcode startup script was fixed so that it would handle 3.x kernels and a missing AMD CPU firmware was added.
Keywords: (none) => validated_updateWhiteboard: MGA2-64-OK MGA2-32-OK feedback => MGA2-64-OK MGA2-32-OK
Update pushed: https://wiki.mageia.org/en/Support/Advisories/MGAA-2012-0154
Status: ASSIGNED => RESOLVEDCC: (none) => tmbResolution: (none) => FIXED