tldr; -- SRC_URI in amd-ucode package is invalid Installing Version 13 AMD64 into system with Athlon II processor, upon checking after first boot for errors, found this for each of the four cpu cores: Feb 2 15:51:54 g1 kernel: microcode: failed to load file amd-ucode/microcode_amd.bin Feb 2 15:51:54 g1 kernel: microcode: CPU1: patch_level=0x010000c8 Researched and found out I needed to install package "amd-ucode" to make the firmware available. Attempted "emerge amd-ucode" Had to add package keyword =sys-firmware/amd-ucode-2012.09.10 ~amd64 -- I realize this may be by design since the status is "unstable" Next, I re-attempted "emerge amd-ucode" and got this: Calculating dependencies... done! >>> Verifying ebuild manifests >>> Emerging (1 of 1) sys-firmware/amd-ucode-2012.09.10 >>> Downloading 'http://www.amd64.org/pub/microcode/amd-ucode-2012-09-10.tar' --2013-02-02 17:40:20-- http://www.amd64.org/pub/microcode/amd-ucode-2012-09-10.tar Resolving www.amd64.org... 217.9.48.20 Connecting to www.amd64.org|217.9.48.20|:80... It appears to me that this line in the ebuild script points to a resource that is no longer available (and hasn't been, for some time): SRC_URI="http://www.amd64.org/pub/microcode/${MY_P}.tar" I manually downloaded the tar file, placed "microcode_amd.bin" into "/lib64/firmware/amd-ucode/" and everything worked: Feb 2 17:09:00 g1 kernel: microcode: CPU1: patch_level=0x010000c8
If you say > It appears to me that this line in the ebuild script points to a resource that is no longer available which I confirm is the case, then how did you manage to do > I manually downloaded the tar file, placed "microcode_amd.bin" into "/lib64/firmware/amd-ucode/" and everything worked then? Did this manual download came from a different URL?
(In reply to comment #1) > If you say > > > It appears to me that this line in the ebuild script points to a resource that is no longer available > > which I confirm is the case, then how did you manage to do > > > I manually downloaded the tar file, placed "microcode_amd.bin" into "/lib64/firmware/amd-ucode/" and everything worked > > then? Did this manual download came from a different URL? Yes, I dug around for a short bit on the AMD site and then just grabbed it from a Fedora source: http://pkgs.fedoraproject.org/repo/pkgs/microcode_ctl/amd-ucode-2012-09-10.tar/559bc355d3799538584add80df2996f0/amd-ucode-2012-09-10.tar
The same Fedora URL was suggested here a few months ago: http://forums.gentoo.org/viewtopic-t-944292-start-0.html And amd64.org is still down. Is it possible to download the tar from that link and make it available in the gentoo mirrors? In this way we'd have a working URL for the ebuild without depending on Fedora.
Can somebody place tar from http://pkgs.fedoraproject.org/repo/pkgs/microcode_ctl/amd-ucode-2012-09-10.tar/559bc355d3799538584add80df2996f0/amd-ucode-2012-09-10.tar to gentoo mirrors?
Is the emerge package really fixed/resolved for the amd64.org site no longer available?... Trying just now to emerge the amd-ucode package gives: # emerge -vDu amd-ucode These are the packages that would be merged, in order: Calculating dependencies... done! [ebuild N ~] sys-firmware/amd-ucode-2012.09.10 70 kB Total: 1 package (1 new), Size of downloads: 70 kB >>> Verifying ebuild manifests >>> Emerging (1 of 1) sys-firmware/amd-ucode-2012.09.10 >>> Downloading 'http://www.amd64.org/pub/microcode/amd-ucode-2012-09-10.tar' --2013-07-29 23:54:44-- http://www.amd64.org/pub/microcode/amd-ucode-2012-09-10.tar Resolving www.amd64.org... 217.9.48.20 Connecting to www.amd64.org|217.9.48.20|:80... failed: Connection timed out. Retrying. --2013-07-29 23:55:46-- (try: 2) http://www.amd64.org/pub/microcode/amd-ucode-2012-09-10.tar Connecting to www.amd64.org|217.9.48.20|:80... failed: Connection timed out. Retrying. --2013-07-29 23:56:48-- (try: 3) http://www.amd64.org/pub/microcode/amd-ucode-2012-09-10.tar Connecting to www.amd64.org|217.9.48.20|:80... failed: Connection timed out. Giving up. !!! Couldn't download 'amd-ucode-2012-09-10.tar'. Aborting. * Fetch failed for 'sys-firmware/amd-ucode-2012.09.10', Log file: * '/var/tmp/portage/sys-firmware/amd-ucode-2012.09.10/temp/build.log' >>> Failed to emerge sys-firmware/amd-ucode-2012.09.10, Log file: >>> '/var/tmp/portage/sys-firmware/amd-ucode-2012.09.10/temp/build.log' * Messages for package sys-firmware/amd-ucode-2012.09.10: * Fetch failed for 'sys-firmware/amd-ucode-2012.09.10', Log file: * '/var/tmp/portage/sys-firmware/amd-ucode-2012.09.10/temp/build.log' Is this easily fixed to use an alternate download? Thanks, Martin
*** This bug has been marked as a duplicate of bug 475380 ***
(In reply to poletti.marco from comment #3) > The same Fedora URL was suggested here a few months ago: > http://forums.gentoo.org/viewtopic-t-944292-start-0.html > > And amd64.org is still down. > > Is it possible to download the tar from that link and make it available in > the gentoo mirrors? In this way we'd have a working URL for the ebuild > without depending on Fedora. If it were possible, it would already have happened sadly. The reason its not already on the gentoo mirrors is this part in the ebuild: > # The license does not allow us to mirror the content. > RESTRICT="mirror" Which basically means upstream have said "If we die, its ok with us that nobody can get this stuff anymore". Silly if you ask me.
*** Bug 475380 has been marked as a duplicate of this bug. ***
We are not allowed to redistribute the tarball so the only viable choice is to rely on the Fedora url (I wonder how come they mirror it themselves)
Did a bit of digging and it seems theres a debian mirror of some kind too, > http://lists.debian.org/debian-release/2012/09/msg00706.html Lead me to > http://anonscm.debian.org/gitweb/?p=users/hmh/amd64-microcode.git;a=tree There's also 3 mirrors listed here: > http://www.filewatcher.com/m/amd-ucode-2012-09-10.tar.71680-0.html Given the circumstances, I'd imagine it pertinent to contact upstream and ask for an exemption that grants us permission to mirror it, given that upstream were last seen at the end of 2012 going "Yeah, the server is down, we're going to get it back at some stage". > https://lkml.org/lkml/2012/12/20/2
Bah, all three mirrors on that filesearch list are dead, so that gives us only the debian git and fedora uri's as valid sources.
Ooh, look, it seems they did get it in linux-firmware! https://git.kernel.org/cgit/linux/kernel/git/firmware/linux-firmware.git/tree/amd-ucode Dated 2013-07-11 And a quick diff of the relevant license in that tree ( https://git.kernel.org/cgit/linux/kernel/git/firmware/linux-firmware.git/tree/LICENSE.amd-ucode ) , suggests they've changed the licensing to a more permissive one.
Hell, we already have an ebuild in tree that solves this problem :D =sys-kernel/linux-firmware-20130728 *BLOCKS* sys-firmware/amd-ucode Its also not mirror restricted. So as soon as that can be stabilized, sys-firmware/amd-ucode can be treecleaned. ( Unless somebody wants to parse the relevant stuff out of linux-firmware into a standalone ebuild )
Now that linux-firmware is getting stable in 485376, it's time to proceed with the removal.
I have one comment/question. I need sys-firmware/amd-ucode for microcode, and can install linux-firmware since: # emerge -pv linux-firmware These are the packages that would be merged, in order: Calculating dependencies... done! [ebuild N ] sys-kernel/linux-firmware-20130728 USE="-savedconfig" 18,793 kB [blocks B ] sys-kernel/linux-firmware[-savedconfig] ("sys-kernel/linux-firmware[-savedconfig]" is blocking x11-drivers/radeon-ucode-20130626) [blocks B ] x11-drivers/radeon-ucode ("x11-drivers/radeon-ucode" is blocking sys-kernel/linux-firmware-20130728) Total: 1 package (1 new), Size of downloads: 18,793 kB Conflict: 2 blocks (2 unsatisfied) And I need x11-drivers/radeon-ucode for: /lib/firmware/radeon/PALM_me.bin /lib/firmware/radeon/PALM_pfp.bin /lib/firmware/radeon/SUMO2_me.bin /lib/firmware/radeon/SUMO2_pfp.bin /lib/firmware/radeon/SUMO_me.bin /lib/firmware/radeon/SUMO_pfp.bin /lib/firmware/radeon/SUMO_rlc.bin /lib/firmware/radeon/SUMO_uvd.bin So what to do?
Uninstall x11-drivers/radeon-ucode and then install sys-kernel/linux-firmware. All the files that you mentioned are also provided by sys-kernel/linux-firmware (this is why they block each other).
@Michael Plichta Easy solution, but lacks a little sophistication which you might want to employ in order to keep radeon-ucode at its most recent, since linux-firmware lacks behind. 1. unmerge radeon-ucode 2. enable the "savedconfig" USE-flag for for linux-firmware 3. emerge linux-firmware 4. move /etc/portage/savedconfig/sys-kernel/linux-firmware-[yourversion] to /etc/portage/savedconfig/sys-kernel/linux-firmware 5. edit said file to not contain any radeon firmware (you might as well remove everything but the amd microcode) 6. emerge linux-firmware again 7. emerge latest radeon-ucode alongside it et voilà - latest radeon microcode + amd microcode available side by side.
(In reply to M. B. from comment #17) > @Michael Plichta > > Easy solution, but lacks a little sophistication which you might want to > employ in order to keep radeon-ucode at its most recent, since > linux-firmware lacks behind. > 1. unmerge radeon-ucode > 2. enable the "savedconfig" USE-flag for for linux-firmware > 3. emerge linux-firmware > 4. move /etc/portage/savedconfig/sys-kernel/linux-firmware-[yourversion] to > /etc/portage/savedconfig/sys-kernel/linux-firmware > 5. edit said file to not contain any radeon firmware (you might as well > remove everything but the amd microcode) > 6. emerge linux-firmware again > 7. emerge latest radeon-ucode alongside it > > et voilà - latest radeon microcode + amd microcode available side by side. I will try it, thx. But which file do you meant in p. 5?
Removed
