Hi! gentoo-sources stable hit the 4.9 branch. However, it breaks the nvidia-drivers when it tries to load it: * Updating module dependencies for 4.9.6-gentoo-r1 ... depmod: WARNING: //lib/modules/4.9.6-gentoo-r1/video/nvidia-uvm.ko needs unknown symbol memset depmod: WARNING: //lib/modules/4.9.6-gentoo-r1/video/nvidia-uvm.ko needs unknown symbol memcpy depmod: WARNING: //lib/modules/4.9.6-gentoo-r1/video/nvidia.ko needs unknown symbol memset After that, it won't load. Then, nvidia optimus does not work. hackenherr:~# dmesg | tail [21:37:27] [ 4650.119349] brcmsmac bcma0:1: brcms_ops_bss_info_changed: qos enabled: false (implement) [ 4650.119358] brcmsmac bcma0:1: brcms_ops_config: change power-save mode: false (implement) [ 4650.119516] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready [ 4966.098835] brcmsmac bcma0:1: brcms_ops_bss_info_changed: qos enabled: false (implement) [ 4966.098843] brcmsmac bcma0:1: brcms_ops_config: change power-save mode: false (implement) [ 4966.098984] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready [ 5282.086799] brcmsmac bcma0:1: brcms_ops_bss_info_changed: qos enabled: false (implement) [ 5282.086807] brcmsmac bcma0:1: brcms_ops_config: change power-save mode: false (implement) [ 5282.086968] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready [ 5462.508988] nvidia: Unknown symbol memset (err 0) Reverting to 4.4 works fine. Reproducible: Always
Did you remember to rebuild the nvidia kernel module after the kernel upgrade? You can use emerge @module-rebuild for this
(In reply to David Hallas from comment #1) > Did you remember to rebuild the nvidia kernel module after the kernel > upgrade? You can use emerge @module-rebuild for this Of course I did, indeed, this report came after I did the rebuild. Tried twice and the same happens. Also,I read elsewhere and I noticed there's some kind of incompatibility between the kernel and the privative driver.
Did you enable: # CONFIG_TRIM_UNUSED_KSYMS is not set ?
(In reply to Sven B. from comment #3) > Did you enable: > # CONFIG_TRIM_UNUSED_KSYMS is not set > ? Nope, doing a grep returned nothing.
Well... I enabled that configuration (CONFIG_TRIM_UNUSED_KSYMS) and this happens: * CONFIG_TRIM_UNUSED_KSYMS: should not be set. But it is. * Please check to make sure these options are set correctly. * Failure to do so may cause unexpected problems. * Once you have satisfied these options, please try merging * this package again. * ERROR: x11-drivers/nvidia-drivers-375.26::gentoo failed (setup phase): * Incorrect kernel configuration options * * Call stack: * ebuild.sh, line 115: Called pkg_setup * nvidia-drivers-375.26.ebuild, line 135: Called linux-mod_pkg_setup * linux-mod.eclass, line 572: Called linux-info_pkg_setup * linux-info.eclass, line 922: Called check_extra_config * linux-info.eclass, line 816: Called die * The specific snippet of code: * die "Incorrect kernel configuration options" * * If you need support, post the output of `emerge --info '=x11-drivers/nvidia-drivers-375.26::gentoo'`, * the complete build log and the output of `emerge -pqv '=x11-drivers/nvidia-drivers-375.26::gentoo'`. * The complete build log is located at '/var/tmp/portage/x11-drivers/nvidia-drivers-375.26/temp/build.log'. * The ebuild environment file is located at '/var/tmp/portage/x11-drivers/nvidia-drivers-375.26/temp/die.env'. * Working directory: '/var/tmp/portage/x11-drivers/nvidia-drivers-375.26/homedir' * S: '/var/tmp/portage/x11-drivers/nvidia-drivers-375.26/work/' >>> Failed to emerge x11-drivers/nvidia-drivers-375.26, Log file: >>> '/var/tmp/portage/x11-drivers/nvidia-drivers-375.26/temp/build.log' * Messages for package x11-drivers/nvidia-drivers-375.26: * CONFIG_TRIM_UNUSED_KSYMS: should not be set. But it is. * Please check to make sure these options are set correctly. * Failure to do so may cause unexpected problems. * Once you have satisfied these options, please try merging * this package again. * ERROR: x11-drivers/nvidia-drivers-375.26::gentoo failed (setup phase): * Incorrect kernel configuration options * * Call stack: * ebuild.sh, line 115: Called pkg_setup * nvidia-drivers-375.26.ebuild, line 135: Called linux-mod_pkg_setup * linux-mod.eclass, line 572: Called linux-info_pkg_setup * linux-info.eclass, line 922: Called check_extra_config * linux-info.eclass, line 816: Called die * The specific snippet of code: * die "Incorrect kernel configuration options" * * If you need support, post the output of `emerge --info '=x11-drivers/nvidia-drivers-375.26::gentoo'`, * the complete build log and the output of `emerge -pqv '=x11-drivers/nvidia-drivers-375.26::gentoo'`. * The complete build log is located at '/var/tmp/portage/x11-drivers/nvidia-drivers-375.26/temp/build.log'. * The ebuild environment file is located at '/var/tmp/portage/x11-drivers/nvidia-drivers-375.26/temp/die.env'. * Working directory: '/var/tmp/portage/x11-drivers/nvidia-drivers-375.26/homedir' * S: '/var/tmp/portage/x11-drivers/nvidia-drivers-375.26/work/'
I had to disable it again (and disable the unused ksyms), and compilation went fine, but with the same errors as the beginning. Kernel >=4.7 and nvidia-drivers may not fit together.
I am using gentoo-sources-4.9.6-r1 and nvidia-drivers-375.26 and it works fine. I do not get any errors when insmodding the drivers. If you want I can attach my kernel config?
(In reply to David Hallas from comment #7) > I am using gentoo-sources-4.9.6-r1 and nvidia-drivers-375.26 and it works > fine. I do not get any errors when insmodding the drivers. If you want I can > attach my kernel config? Well.. it may be a good idea. Also, I patched the kernel to build with ld.gold, I don't know if that's the culprit.
Created attachment 462796 [details] Working kernel config
I think it is a good idea to re-compile the kernel with the GNU linker to test. I have experienced similar problems when mixing libraries build with Gold and the GNU linker.
(In reply to David Hallas from comment #10) > I think it is a good idea to re-compile the kernel with the GNU linker to > test. I have experienced similar problems when mixing libraries build with > Gold and the GNU linker. I'm working on my memoir about Clang/LLVM and it recommends to use the Gold linker. I think it's better to make the kernel building to force to use the default linker instead.
Created attachment 462802 [details] Xorg.0.log
Created attachment 462804 [details] dmesg output
I am having the same problem with sys-kernel/gentoo-sources-4.9.6-r1 compiled with genkernel. Above i just uploaded Xorg.0.log and the dmesg output. Upon the end of the booting process, sddm crashes with this error message: login: sddm[5272]: segfault at 7c0 ip 00000000000007c0 sp 00007ffea701a638 error 14 in sddm[400000+6d000] The bug probably concerns only older "legacy" nVidia cards such as my GeForce 8600 GT: https://www.nvidia.com/object/IO_32667.html This probably explains why David Hallas did not experience the bug with his newer driver, in comment #7 . Look also here for more information, and a possible patch: https://bbs.archlinux.org/viewtopic.php?id=222417
Sorry, i forgot to mention my nvidia driver: x11-drivers/nvidia-drivers-340.101-r1:0/340
Well, my Nvidia Card is a GeForce GT 540M, which is not legacy. Also, I compiled the kernel with genkernel. As a reference.
issues with legacy cards are in another bug report with the patch: https://bugs.gentoo.org/show_bug.cgi?id=602844
Thanks Gleb, sorry about the noise in here... I don't see how i can delete my attachments, so i marked them as obsolete... Looks like my bug has been resolved in the other thread, best of luck for this one, too... ;-)
I found that my video card (Nvidia GeForce GT 540M) is now a legacy card. Tried with 340.xx version and working fine. (Also I found a patch that I'll submit then) (In reply to David Hallas from comment #7) > I am using gentoo-sources-4.9.6-r1 and nvidia-drivers-375.26 and it works > fine. I do not get any errors when insmodding the drivers. If you want I can > attach my kernel config? Just in case, what's your card model?
Created attachment 463074 [details, diff] The patch for the legacy drivers This patch is working. Also, as a suggestion, the [compat] as a dependency in the x11-misc/primus package is not needed because it works fine without the [compat] feature.
(In reply to jorgicio from comment #19) > I found that my video card (Nvidia GeForce GT 540M) is now a legacy card. > Tried with 340.xx version and working fine. (Also I found a patch that I'll > submit then) > > (In reply to David Hallas from comment #7) > > I am using gentoo-sources-4.9.6-r1 and nvidia-drivers-375.26 and it works > > fine. I do not get any errors when insmodding the drivers. If you want I can > > attach my kernel config? > > Just in case, what's your card model? My card is a NVIDIA GPU GeForce 940MX (GM108-A)
GT 540M is definitely not a legacy card, it's a Fermi based chip so it's relatevely new. At least it is not listed nowhere on these pages: https://nvidia.custhelp.com/app/answers/detail/a_id/3142 https://www.nvidia.com/object/IO_32667.html I may have missed some news sources, though.
(In reply to Gleb from comment #22) > GT 540M is definitely not a legacy card, it's a Fermi based chip so it's > relatevely new. At least it is not listed nowhere on these pages: > > https://nvidia.custhelp.com/app/answers/detail/a_id/3142 > https://www.nvidia.com/object/IO_32667.html > > I may have missed some news sources, though. Well... I thought so, but this page gets me confused: http://www.nvidia.com/page/legacy.html Also, with 340.xx, the card works fine, but not with recent versions (tried with 370.xx too)
(In reply to David Hallas from comment #21) > (In reply to jorgicio from comment #19) > > I found that my video card (Nvidia GeForce GT 540M) is now a legacy card. > > Tried with 340.xx version and working fine. (Also I found a patch that I'll > > submit then) > > > > (In reply to David Hallas from comment #7) > > > I am using gentoo-sources-4.9.6-r1 and nvidia-drivers-375.26 and it works > > > fine. I do not get any errors when insmodding the drivers. If you want I can > > > attach my kernel config? > > > > Just in case, what's your card model? > > My card is a NVIDIA GPU GeForce 940MX (GM108-A) Definitively your card is not legacy, so that's may explain why you don't have this problem. Thanks.
this legacy driver patch is added and is solved in >4.9.6 and we cannot support tainted kernel with propietary module *** This bug has been marked as a duplicate of bug 602844 ***
