[gentoo-user] Hardened Kernel and UKSM
Hi, I'm trying to patch Gentoo Hardened sources 3.15.5r2 with UKSM. It patched fine, got some rejects on fork.c, mmap.c and exec.c. I saw the code, they were trivial so I added the changes manually. But I need help on this one: In file included from include/linux/ksm.h:138:0, from kernel/fork.c:56: include/linux/uksm.h: In function ‘uksm_cow_page’: include/linux/uksm.h:74:9: error: ‘struct vm_area_struct’ has no member named ‘uksm_vma_slot’ include/linux/uksm.h:75:6: error: ‘struct vm_area_struct’ has no member named ‘uksm_vma_slot’ include/linux/uksm.h: In function ‘uksm_cow_pte’: include/linux/uksm.h:80:9: error: ‘struct vm_area_struct’ has no member named ‘uksm_vma_slot’ include/linux/uksm.h:81:6: error: ‘struct vm_area_struct’ has no member named ‘uksm_vma_slot’ make[1]: *** [kernel/fork.o] Error 1 make: *** [kernel] Error 2 -- Nilesh Govindrajan m...@nileshgr.com
[gentoo-user] Re: Hardened Kernel and UKSM
On Sat, Aug 23, 2014 at 12:03 PM, Nilesh Govindrajan m...@nileshgr.com wrote: Hi, I'm trying to patch Gentoo Hardened sources 3.15.5r2 with UKSM. It patched fine, got some rejects on fork.c, mmap.c and exec.c. I saw the code, they were trivial so I added the changes manually. But I need help on this one: In file included from include/linux/ksm.h:138:0, from kernel/fork.c:56: include/linux/uksm.h: In function ‘uksm_cow_page’: include/linux/uksm.h:74:9: error: ‘struct vm_area_struct’ has no member named ‘uksm_vma_slot’ include/linux/uksm.h:75:6: error: ‘struct vm_area_struct’ has no member named ‘uksm_vma_slot’ include/linux/uksm.h: In function ‘uksm_cow_pte’: include/linux/uksm.h:80:9: error: ‘struct vm_area_struct’ has no member named ‘uksm_vma_slot’ include/linux/uksm.h:81:6: error: ‘struct vm_area_struct’ has no member named ‘uksm_vma_slot’ make[1]: *** [kernel/fork.o] Error 1 make: *** [kernel] Error 2 Ouch, just after writing to the list I managed to crack it. It turns out that the piece of code which defined the structure member got into some other struct instead of vm_area_struct. Anyway, I'd like to know if anybody has tried UKSM on hardened kernel and if there are any problems, etc
Re: [gentoo-user] re: sys-fs/lvm2 question
On 08/22/2014 10:36 PM, Canek Peláez Valdés wrote: On Fri, Aug 22, 2014 at 1:56 PM, Alexander Kapshuk alexander.kaps...@gmail.com wrote: As I updated my system today, I noticed that 'sys-fs/lvm2' got updated amongst other packages as well. I don't use LVM on my system. If I understand it correctly, 'sys-fs/lvm2' is a required dependency for 'sys-fs/udisks/udisks-1.0.5-r1': equery d sys-fs/lvm2 * These packages depend on sys-fs/lvm2: sys-block/parted-3.1-r1 (device-mapper ? =sys-fs/lvm2-2.02.45) sys-boot/grub-2.00_p5107-r2 (device-mapper ? =sys-fs/lvm2-2.02.45) sys-fs/udisks-1.0.5-r1 (=sys-fs/lvm2-2.02.66) sys-fs/udisks-2.1.3 (cryptsetup ? sys-fs/lvm2[udev(+)]) equery -q u sys-block/parted | grep device-mapper -device-mapper equery -q u sys-boot/grub | grep device-mapper -device-mapper equery -q u '=sys-fs/udisks-1.0.5-r1' -debug +nls -remote-access $ equery -q u '=sys-fs/udisks-2.1.3' | grep cryptsetup -cryptsetup /usr/portage/sys-fs/udisks/udisks-1.0.5-r1.ebuild:17,24 COMMON_DEPEND==dev-libs/dbus-glib-0.100 snip =sys-fs/lvm2-2.02.66 What are my options, if I were to remove 'sys-fs/lvm2' altogether? Remove sys-fs/udisks:0, which depends unconditionally on LVM2; also, it's on life support, AFAIR. sys-fs/udisks:2 is actively maintained and it depends only conditionally on LVM2. What would you recommend doing about it? What does depend on sys-fs/udisks? What's the output from equery d sys-fs/udisks? Most applications switched to udisks-2, but some are still stuck with udisks-1 (XMBC, now Kodi, comes to mind). If an application that you absolutely need requires sys-fs/udisks:0, then you will need LVM2 also. Regards. Looks like I've got a couple of apps that do require udisks-1 to run: equery d sys-fs/udisks * These packages depend on sys-fs/udisks: gnome-base/gvfs-1.20.2 (udisks ? =sys-fs/udisks-1.97:2) xfce-extra/xfce4-power-manager-1.3.0 (udisks ? sys-fs/udisks:0) So I'm going to have to keep lvm2 then. Thanks.
Re: [gentoo-user] re: sys-fs/lvm2 question
On Sat, Aug 23, 2014 at 1:47 AM, Alexander Kapshuk alexander.kaps...@gmail.com wrote: On 08/22/2014 10:36 PM, Canek Peláez Valdés wrote: On Fri, Aug 22, 2014 at 1:56 PM, Alexander Kapshuk alexander.kaps...@gmail.com wrote: As I updated my system today, I noticed that 'sys-fs/lvm2' got updated amongst other packages as well. I don't use LVM on my system. If I understand it correctly, 'sys-fs/lvm2' is a required dependency for 'sys-fs/udisks/udisks-1.0.5-r1': equery d sys-fs/lvm2 * These packages depend on sys-fs/lvm2: sys-block/parted-3.1-r1 (device-mapper ? =sys-fs/lvm2-2.02.45) sys-boot/grub-2.00_p5107-r2 (device-mapper ? =sys-fs/lvm2-2.02.45) sys-fs/udisks-1.0.5-r1 (=sys-fs/lvm2-2.02.66) sys-fs/udisks-2.1.3 (cryptsetup ? sys-fs/lvm2[udev(+)]) equery -q u sys-block/parted | grep device-mapper -device-mapper equery -q u sys-boot/grub | grep device-mapper -device-mapper equery -q u '=sys-fs/udisks-1.0.5-r1' -debug +nls -remote-access $ equery -q u '=sys-fs/udisks-2.1.3' | grep cryptsetup -cryptsetup /usr/portage/sys-fs/udisks/udisks-1.0.5-r1.ebuild:17,24 COMMON_DEPEND==dev-libs/dbus-glib-0.100 snip =sys-fs/lvm2-2.02.66 What are my options, if I were to remove 'sys-fs/lvm2' altogether? Remove sys-fs/udisks:0, which depends unconditionally on LVM2; also, it's on life support, AFAIR. sys-fs/udisks:2 is actively maintained and it depends only conditionally on LVM2. What would you recommend doing about it? What does depend on sys-fs/udisks? What's the output from equery d sys-fs/udisks? Most applications switched to udisks-2, but some are still stuck with udisks-1 (XMBC, now Kodi, comes to mind). If an application that you absolutely need requires sys-fs/udisks:0, then you will need LVM2 also. Regards. Looks like I've got a couple of apps that do require udisks-1 to run: equery d sys-fs/udisks * These packages depend on sys-fs/udisks: gnome-base/gvfs-1.20.2 (udisks ? =sys-fs/udisks-1.97:2) gvfs depends on sys-fs/udisk:2, so this one doesn't need udisks-1. xfce-extra/xfce4-power-manager-1.3.0 (udisks ? sys-fs/udisks:0) What does xfce4-power-manager uses udisks for? You could try to emerge it with USE=-udisks and see if you miss some functionality. If you don't, you can get rid of udisks-1 and LVM2. Regards. -- Canek Peláez Valdés Profesor de asignatura, Facultad de Ciencias Universidad Nacional Autónoma de México
Re: [gentoo-user] re: sys-fs/lvm2 question
On 08/23/2014 09:53 AM, Canek Peláez Valdés wrote: On Sat, Aug 23, 2014 at 1:47 AM, Alexander Kapshuk alexander.kaps...@gmail.com wrote: On 08/22/2014 10:36 PM, Canek Peláez Valdés wrote: On Fri, Aug 22, 2014 at 1:56 PM, Alexander Kapshuk alexander.kaps...@gmail.com wrote: As I updated my system today, I noticed that 'sys-fs/lvm2' got updated amongst other packages as well. I don't use LVM on my system. If I understand it correctly, 'sys-fs/lvm2' is a required dependency for 'sys-fs/udisks/udisks-1.0.5-r1': equery d sys-fs/lvm2 * These packages depend on sys-fs/lvm2: sys-block/parted-3.1-r1 (device-mapper ? =sys-fs/lvm2-2.02.45) sys-boot/grub-2.00_p5107-r2 (device-mapper ? =sys-fs/lvm2-2.02.45) sys-fs/udisks-1.0.5-r1 (=sys-fs/lvm2-2.02.66) sys-fs/udisks-2.1.3 (cryptsetup ? sys-fs/lvm2[udev(+)]) equery -q u sys-block/parted | grep device-mapper -device-mapper equery -q u sys-boot/grub | grep device-mapper -device-mapper equery -q u '=sys-fs/udisks-1.0.5-r1' -debug +nls -remote-access $ equery -q u '=sys-fs/udisks-2.1.3' | grep cryptsetup -cryptsetup /usr/portage/sys-fs/udisks/udisks-1.0.5-r1.ebuild:17,24 COMMON_DEPEND==dev-libs/dbus-glib-0.100 snip =sys-fs/lvm2-2.02.66 What are my options, if I were to remove 'sys-fs/lvm2' altogether? Remove sys-fs/udisks:0, which depends unconditionally on LVM2; also, it's on life support, AFAIR. sys-fs/udisks:2 is actively maintained and it depends only conditionally on LVM2. What would you recommend doing about it? What does depend on sys-fs/udisks? What's the output from equery d sys-fs/udisks? Most applications switched to udisks-2, but some are still stuck with udisks-1 (XMBC, now Kodi, comes to mind). If an application that you absolutely need requires sys-fs/udisks:0, then you will need LVM2 also. Regards. Looks like I've got a couple of apps that do require udisks-1 to run: equery d sys-fs/udisks * These packages depend on sys-fs/udisks: gnome-base/gvfs-1.20.2 (udisks ? =sys-fs/udisks-1.97:2) gvfs depends on sys-fs/udisk:2, so this one doesn't need udisks-1. xfce-extra/xfce4-power-manager-1.3.0 (udisks ? sys-fs/udisks:0) What does xfce4-power-manager uses udisks for? You could try to emerge it with USE=-udisks and see if you miss some functionality. If you don't, you can get rid of udisks-1 and LVM2. Regards. Thanks for pointing that out. I overlooked that. I'll give that a try.
Re: [gentoo-user] re: sys-fs/lvm2 question
On 23/08/14 09:53, Canek Peláez Valdés wrote: On Sat, Aug 23, 2014 at 1:47 AM, Alexander Kapshuk alexander.kaps...@gmail.com wrote: On 08/22/2014 10:36 PM, Canek Peláez Valdés wrote: On Fri, Aug 22, 2014 at 1:56 PM, Alexander Kapshuk alexander.kaps...@gmail.com wrote: As I updated my system today, I noticed that 'sys-fs/lvm2' got updated amongst other packages as well. I don't use LVM on my system. If I understand it correctly, 'sys-fs/lvm2' is a required dependency for 'sys-fs/udisks/udisks-1.0.5-r1': equery d sys-fs/lvm2 * These packages depend on sys-fs/lvm2: sys-block/parted-3.1-r1 (device-mapper ? =sys-fs/lvm2-2.02.45) sys-boot/grub-2.00_p5107-r2 (device-mapper ? =sys-fs/lvm2-2.02.45) sys-fs/udisks-1.0.5-r1 (=sys-fs/lvm2-2.02.66) sys-fs/udisks-2.1.3 (cryptsetup ? sys-fs/lvm2[udev(+)]) equery -q u sys-block/parted | grep device-mapper -device-mapper equery -q u sys-boot/grub | grep device-mapper -device-mapper equery -q u '=sys-fs/udisks-1.0.5-r1' -debug +nls -remote-access $ equery -q u '=sys-fs/udisks-2.1.3' | grep cryptsetup -cryptsetup /usr/portage/sys-fs/udisks/udisks-1.0.5-r1.ebuild:17,24 COMMON_DEPEND==dev-libs/dbus-glib-0.100 snip =sys-fs/lvm2-2.02.66 What are my options, if I were to remove 'sys-fs/lvm2' altogether? Remove sys-fs/udisks:0, which depends unconditionally on LVM2; also, it's on life support, AFAIR. sys-fs/udisks:2 is actively maintained and it depends only conditionally on LVM2. What would you recommend doing about it? What does depend on sys-fs/udisks? What's the output from equery d sys-fs/udisks? Most applications switched to udisks-2, but some are still stuck with udisks-1 (XMBC, now Kodi, comes to mind). If an application that you absolutely need requires sys-fs/udisks:0, then you will need LVM2 also. Regards. Looks like I've got a couple of apps that do require udisks-1 to run: equery d sys-fs/udisks * These packages depend on sys-fs/udisks: gnome-base/gvfs-1.20.2 (udisks ? =sys-fs/udisks-1.97:2) gvfs depends on sys-fs/udisk:2, so this one doesn't need udisks-1. xfce-extra/xfce4-power-manager-1.3.0 (udisks ? sys-fs/udisks:0) What does xfce4-power-manager uses udisks for? You could try to emerge it with USE=-udisks and see if you miss some functionality. If you don't, you can get rid of udisks-1 and LVM2. Regards. xfce4-power-manager-1.3.0 and older uses UDisks 1.x for controlling disk spinning, like to reduce it xfce4-power-manager-1.3.1 and higher removed UDisks 1.x dependency and the spindown feature, supposedly it had issues and doesn't work with SSD anyway... anyways, upstream decision to not use udisks anymore so, i recommend upgrading to 1.3.1, adding it to package.keywords if required thanks, samuli
Re: [gentoo-user] re: sys-fs/lvm2 question
On 08/23/2014 09:53 AM, Canek Peláez Valdés wrote: On Sat, Aug 23, 2014 at 1:47 AM, Alexander Kapshuk alexander.kaps...@gmail.com wrote: On 08/22/2014 10:36 PM, Canek Peláez Valdés wrote: On Fri, Aug 22, 2014 at 1:56 PM, Alexander Kapshuk alexander.kaps...@gmail.com wrote: As I updated my system today, I noticed that 'sys-fs/lvm2' got updated amongst other packages as well. I don't use LVM on my system. If I understand it correctly, 'sys-fs/lvm2' is a required dependency for 'sys-fs/udisks/udisks-1.0.5-r1': equery d sys-fs/lvm2 * These packages depend on sys-fs/lvm2: sys-block/parted-3.1-r1 (device-mapper ? =sys-fs/lvm2-2.02.45) sys-boot/grub-2.00_p5107-r2 (device-mapper ? =sys-fs/lvm2-2.02.45) sys-fs/udisks-1.0.5-r1 (=sys-fs/lvm2-2.02.66) sys-fs/udisks-2.1.3 (cryptsetup ? sys-fs/lvm2[udev(+)]) equery -q u sys-block/parted | grep device-mapper -device-mapper equery -q u sys-boot/grub | grep device-mapper -device-mapper equery -q u '=sys-fs/udisks-1.0.5-r1' -debug +nls -remote-access $ equery -q u '=sys-fs/udisks-2.1.3' | grep cryptsetup -cryptsetup /usr/portage/sys-fs/udisks/udisks-1.0.5-r1.ebuild:17,24 COMMON_DEPEND==dev-libs/dbus-glib-0.100 snip =sys-fs/lvm2-2.02.66 What are my options, if I were to remove 'sys-fs/lvm2' altogether? Remove sys-fs/udisks:0, which depends unconditionally on LVM2; also, it's on life support, AFAIR. sys-fs/udisks:2 is actively maintained and it depends only conditionally on LVM2. What would you recommend doing about it? What does depend on sys-fs/udisks? What's the output from equery d sys-fs/udisks? Most applications switched to udisks-2, but some are still stuck with udisks-1 (XMBC, now Kodi, comes to mind). If an application that you absolutely need requires sys-fs/udisks:0, then you will need LVM2 also. Regards. Looks like I've got a couple of apps that do require udisks-1 to run: equery d sys-fs/udisks * These packages depend on sys-fs/udisks: gnome-base/gvfs-1.20.2 (udisks ? =sys-fs/udisks-1.97:2) gvfs depends on sys-fs/udisk:2, so this one doesn't need udisks-1. xfce-extra/xfce4-power-manager-1.3.0 (udisks ? sys-fs/udisks:0) What does xfce4-power-manager uses udisks for? You could try to emerge it with USE=-udisks and see if you miss some functionality. If you don't, you can get rid of udisks-1 and LVM2. Regards. I've put xfce-extra/xfce4-power-manager -udisk into /etc/portage/package.use. Ran emerge -avuND @world, which only reinstalled the xfce power mananger with udisks disabled. emerge --depclean suggested removing the apps listed below, which happened to be all interdependent: sys-block/thin-provisioning-tools-0.3.2-r1 sys-fs/lvm2-2.02.109 sys-fs/udisks-1.0.5-r1 sys-apps/rescan-scsi-bus-1.29 sys-apps/sg3_utils-1.37 equery d sys-block/thin-provisioning-tools * These packages depend on sys-block/thin-provisioning-tools: sys-fs/lvm2-2.02.109 (thin ? =sys-block/thin-provisioning-tools-0.3.0) $ equery d sys-apps/rescan-scsi-bus * These packages depend on sys-apps/rescan-scsi-bus: sys-apps/sg3_utils-1.37 (=sys-apps/rescan-scsi-bus-1.24) $ equery d sys-apps/sg3_utils * These packages depend on sys-apps/sg3_utils: sys-apps/rescan-scsi-bus-1.29 (=sys-apps/sg3_utils-1.24) sys-fs/udisks-1.0.5-r1 (=sys-apps/sg3_utils-1.27.20090411) So I had portage remove those. The Linux from Scratch page for xfce4-power-manger 1.2.0, http://www.linuxfromscratch.org/blfs/view/svn/xfce/xfce4-power-manager.html, lists udisks as an optional dependency. Thanks very much for your help.
Re: [gentoo-user] re: sys-fs/lvm2 question
On 08/23/2014 10:31 AM, Samuli Suominen wrote: xfce4-power-manager-1.3.0 and older uses UDisks 1.x for controlling disk spinning, like to reduce it xfce4-power-manager-1.3.1 and higher removed UDisks 1.x dependency and the spindown feature, supposedly it had issues and doesn't work with SSD anyway... anyways, upstream decision to not use udisks anymore so, i recommend upgrading to 1.3.1, adding it to package.keywords if required thanks, samuli Thanks for your response. I remember being advised on this list against mixing both stable and unstable packages as much as possible. Does that still hold true? Or would it be OK to pull this one in without braking anything unnecessarily? Thanks.
Re: [gentoo-user] re: sys-fs/lvm2 question
On 23/08/2014 09:51, Alexander Kapshuk wrote: On 08/23/2014 10:31 AM, Samuli Suominen wrote: xfce4-power-manager-1.3.0 and older uses UDisks 1.x for controlling disk spinning, like to reduce it xfce4-power-manager-1.3.1 and higher removed UDisks 1.x dependency and the spindown feature, supposedly it had issues and doesn't work with SSD anyway... anyways, upstream decision to not use udisks anymore so, i recommend upgrading to 1.3.1, adding it to package.keywords if required thanks, samuli Thanks for your response. I remember being advised on this list against mixing both stable and unstable packages as much as possible. Does that still hold true? Or would it be OK to pull this one in without braking anything unnecessarily? I think you have a wrong impression. There is actually not much wrong with mixing stable and unstable as long as you do it sensibly. What you shouldn't do is to wantonly mix packages in @stable and other basic libs and still expect it to work. Stable gcc and unstable glibc with jpeg, zlib and openssl all mixed and matched any old way is certain to show inconsistencies (as you will be the only person who has ever tested that combination). What is being proposed here is that you take one userland package (xfce4-power-manager) and upgrade it to the new version. It's highly unlikely to break anything and I can tell that just by looking at it's purpose and where it fits in the stack. It will either work or not, and the list of things that might link to it are a rather small list indeed. So just give it a spin, you can always revert if it's incompatible with everything else you have. The answer to the last question you pose is correctly mu as no-one can possibly answer it properly. The best we can do for you is paint the big picture and ask you to try then report back if it works, as I have done above. -- Alan McKinnon alan.mckin...@gmail.com
Re: [gentoo-user] re: sys-fs/lvm2 question
On 08/23/2014 11:22 AM, Alan McKinnon wrote: On 23/08/2014 09:51, Alexander Kapshuk wrote: On 08/23/2014 10:31 AM, Samuli Suominen wrote: xfce4-power-manager-1.3.0 and older uses UDisks 1.x for controlling disk spinning, like to reduce it xfce4-power-manager-1.3.1 and higher removed UDisks 1.x dependency and the spindown feature, supposedly it had issues and doesn't work with SSD anyway... anyways, upstream decision to not use udisks anymore so, i recommend upgrading to 1.3.1, adding it to package.keywords if required thanks, samuli Thanks for your response. I remember being advised on this list against mixing both stable and unstable packages as much as possible. Does that still hold true? Or would it be OK to pull this one in without braking anything unnecessarily? I think you have a wrong impression. There is actually not much wrong with mixing stable and unstable as long as you do it sensibly. What you shouldn't do is to wantonly mix packages in @stable and other basic libs and still expect it to work. Stable gcc and unstable glibc with jpeg, zlib and openssl all mixed and matched any old way is certain to show inconsistencies (as you will be the only person who has ever tested that combination). What is being proposed here is that you take one userland package (xfce4-power-manager) and upgrade it to the new version. It's highly unlikely to break anything and I can tell that just by looking at it's purpose and where it fits in the stack. It will either work or not, and the list of things that might link to it are a rather small list indeed. So just give it a spin, you can always revert if it's incompatible with everything else you have. The answer to the last question you pose is correctly mu as no-one can possibly answer it properly. The best we can do for you is paint the big picture and ask you to try then report back if it works, as I have done above. I'll give that whirl. Thanks.
[gentoo-user] Pin a package to a binary (quickpkg'd) version?
Is it possible to do this? Thanks...
[gentoo-user] Re:(Bug 510912) mesos ebuild
James wireless at tampabay.rr.com writes: mesos-0.18.0-r1.ebuild OOps, forget to post the link. http://gpo.zugaina.org/sys-cluster/mesos/RDep James
Re: [gentoo-user] Pin a package to a binary (quickpkg'd) version?
On 23/08/2014 12:34, Tanstaafl wrote: Is it possible to do this? Thanks... Not directly. I'm assuming you mean packages you built yourself and quick-pkg'ed them, not something available as a -bin You can use emerge -K, so emerge will fail if there's no binpkg available. This will do what you want as long as you a) always use the -K option b) don't try emerge something else as well Portage is designed to build your packages from source; binpkgs are very much a third class citizen with only very primitive levels of support. -- Alan McKinnon alan.mckin...@gmail.com
Re: [gentoo-user] Pin a package to a binary (quickpkg'd) version?
On 8/23/2014 8:16 AM, Alan McKinnon alan.mckin...@gmail.com wrote: On 23/08/2014 12:34, Tanstaafl wrote: Is it possible to do this? Not directly. I'm assuming you mean packages you built yourself and quick-pkg'ed them, not something available as a -bin Correct... I have buildpkg feature enabled in make.conf, so everything gets quickpkg'd... You can use emerge -K, so emerge will fail if there's no binpkg available. This will do what you want as long as you a) always use the -K option b) don't try emerge something else as well Portage is designed to build your packages from source; binpkgs are very much a third class citizen with only very primitive levels of support. Bummer... What I want is to be able to pin a specific package to the quickpkg'd version, so it doesn't get updated during an emerge world...
Re: [gentoo-user] re: trouble with USB mouse on laptop
On 08/08/2014 10:46 AM, J. Roeleveld wrote: On Friday, August 08, 2014 10:30:55 AM Alexander Kapshuk wrote: Howdy, I've been having trouble with USB mice on my laptop, with the mouse pointer and the scroll wheel not responding all the time. It's not until I click either button that the mouse starts responding again. The output taken from /var/log/messages shown below does not seem out of the ordinary: Aug 8 09:32:56 localhost kernel: [ 5692.801085] usb 4-2: new low-speed USB device number 6 using uhci_h cd Aug 8 09:32:56 localhost kernel: [ 5692.968100] usb 4-2: New USB device found, idVendor=045e, idProduct =0084 Aug 8 09:32:56 localhost kernel: [ 5692.968109] usb 4-2: New USB device strings: Mfr=1, Product=2, Seri alNumber=0 Aug 8 09:32:56 localhost kernel: [ 5692.968116] usb 4-2: Product: Microsoft Basic Optical Mouse Aug 8 09:32:56 localhost kernel: [ 5692.968121] usb 4-2: Manufacturer: Microsoft Aug 8 09:32:56 localhost kernel: [ 5692.995929] input: Microsoft Microsoft Basic Optical Mouse as /devi ces/pci:00/:00:1a.1/usb4/4-2/4-2:1.0/0003:045E:0084.0005/input/input 12 Aug 8 09:32:56 localhost kernel: [ 5692.996296] hid-generic 0003:045E:0084.0005: input,hidraw0: USB HID v1.11 Mouse [Microsoft Microsoft Basic Optical Mouse] on usb-:00:1a.1-2/input0 Aug 8 09:32:56 localhost mtp-probe: checking bus 4, device 6: /sys/devices/pci:00/:00:1a.1/usb4/4-2 Aug 8 09:32:56 localhost mtp-probe: bus: 4, device: 6 was not an MTP device Your advice would be appreciated. Alexander Kapshuk. Looks like quite a few people have issues with that mouse. Googling for idVendor=045e, idProduct=0084 brings up several hits with people describing what you are seeing. Basically, it looks like the latest firmware on the mouse seems to have some powersaving functions that don't work properly. Can you test with a different mouse to see if that one works? Alternatively, disable USB powersaving: # for i in /sys/bus/usb/devices/*/power/autosuspend; do echo 2 $i; done # for foo in /sys/bus/usb/devices/*/power/level; do echo on $foo; done -- Joost I've done as suggested above as well as set AUTOSUSPEND_USBID_BLACKLIST in /etc/laptop-mode/conf.d/usb-autosuspend.conf to the ID of my USB mouse. So far the mouse has been working OK. Thanks for your help.
Re: [gentoo-user] re: automounting removable drives
On 10/08/2013 09:21 PM, Samuli Suominen wrote: futhermore authorization from polkit/consolekit must be working, so you must see 'active = TRUE' line when you run `ck-list-sessions` in your Xfce's Terminal as a normal user, see this thread (first post of it): http://forums.gentoo.org/viewtopic-t-858965-start-0.html and like said, xfce4-mount-plugin is irrelevant, and `mount` command shouldn't be used at all for udisks maintained removable devices, instead `udisksctl mount` should be used as a normal user if you really want to mount from commandline I followed the instructions given above to set up for my removable drives to automount on my laptop and desktop machines. I got it to work flawlessly on my laptop, but not on the desktop. What I seem to be having trouble figuring out is why my DVDs won't get automounted if I eject a DVD and then load it again. The DVDs only seem to get automounted when I first boot my system. I've checked the system logs, compared the recommended settings given at the gentoo forum link above against the settings on my laptop. Still didn't find anything amiss. I little help would be appreciated. Here's some info that might prove handy. dmesg output: [5.076608] ata5.00: ATAPI: Optiarc DVD RW AD-5200A, 1.01, max UDMA/66 [5.082537] ata5.00: configured for UDMA/66 [5.085298] scsi 4:0:0:0: CD-ROMOptiarc DVD RW AD-5200A 1.01 PQ: 0 ANSI: 5 [5.087417] sr0: scsi3-mmc drive: 94x/94x writer cd/rw xa/form2 cdda tray [5.087421] cdrom: Uniform CD-ROM driver Revision: 3.20 [5.087668] sr 4:0:0:0: Attached scsi CD-ROM sr0 [5.087886] sr 4:0:0:0: Attached scsi generic sg1 type 5 udisksctl status MODEL REVISION SERIAL DEVICE -- ST3320613AS SD11 9SZ08MMB sda Optiarc DVD RW AD-5200A 1.01 Optiarc_DVD_RW_AD-5200A sr0 /var/log/messages showing the udisks activities when I unmounted the DVD that got mounted at boot time. Aug 23 16:32:00 localhost kernel: [24817.787649] UDF-fs: INFO Mounting volume 'Adventures of T', timestamp 2012/12/17 20:58 (10b4) Aug 23 16:32:00 localhost udisksd[3035]: Mounted /dev/sr0 at /run/media/sasha/ADVENTURES OF TINTIN on behalf of uid 1000 Aug 23 16:32:07 localhost udisksd[3035]: Cleaning up mount point /run/media/sasha/ADVENTURES OF TINTIN (device 11:0 is not mounted) Aug 23 16:32:07 localhost udisksd[3035]: Unmounted /dev/sr0 on behalf of uid 1000
[gentoo-user] Re: Pin a package to a binary (quickpkg'd) version?
Tanstaafl wrote: Bummer... What I want is to be able to pin a specific package to the quickpkg'd version, so it doesn't get updated during an emerge world... Wouldn't it mostly work if you a] Copied the package ebuild directory to a local overlay to ensure the version keeps being available to portage even if it vanishes from the official tree and b] masked any later version in, say, /etc/portage/package.mask/pinned ( and c] mentally prepared to repeat the procedure when the tree no longer provides compatible dependencies ) ?
[gentoo-user] Re: Pin a package to a binary (quickpkg'd) version?
Jouni Kosonen jouni.kosonen at tukesoft.com writes: Tanstaafl wrote: What I want is to be able to pin a specific package to the quickpkg'd version, so it doesn't get updated during an emerge world... Wouldn't it mostly work if you a] Copied the package ebuild directory to a local overlay to ensure the version keeps being available to portage even if it vanishes from the official tree and b] masked any later version in, say, /etc/portage/package.mask/pinned ( and c] mentally prepared to repeat the procedure when the tree no longer provides compatible dependencies ) Very similar to what I was musing/researching Some resources you may want to read up on that deal with slotting, and multislots (which allow for multiple versions (concurrent) installations for certain key packages. http://devmanual.gentoo.org/general-concepts/slotting/ euse -i multislot http://www.gnu.org/software/autoconf/autoconf.html A recent post on gentoo-dev distutils-r1: set install paths via setup.cfg rather than argv Just might provide you some ideas how to achieve what you want to do. I'm looking at a similar issue for my local overlays and codes I'm testing in a variety of forms; ebuilds and raw sources. hth, James
[gentoo-user] PAM auto-adding .db suffix: feature or bug?
Hi Gentoo-users, after playing with PAM for a few days I discovered one strange thing: PAM adds .db suffix to database definition on its own! What I mean is following: _ more /etc/pam.d/postfix authrequired pam_userdb.so db=/etc/postfix/virtual_users.db account required pam_userdb.so db=/etc/postfix/virtual_users.db ls -l /etc/postfix/virtual_users.db -rw--- 1 root root 12288 Aug 20 18:45 /etc/postfix/virtual_users.db testsaslauthd -u user1 -p password1 -s postfix 0: NO authentication failed tail -n1 saslauth.log saslauthd: pam_userdb(postfix:auth): user_lookup: could not open database '/etc/postfix/virtual_users.db': no such file or directory __ Then I changed /etc/pam.d/postfix to: authrequired pam_userdb.so db=/etc/postfix/virtual_users account required pam_userdb.so db=/etc/postfix/virtual_users And now everything works. But there is no /etc/postfix/virtual_users file, only /etc/postfix/virtual_users.db! So it seems to me PAM adds .db to database-definition itself and user has to define it without .db, otherwise what PAM is actually looking for is /etc/postfix/virtual_users.db.db Is this normal? Never seen this Windows-like behaviour with auto-adding suffix in *nix world. And I did not find anything about it in PAM-documentation. It says just: pam_userdb.so db=/path/database Not a single word that user has to cut .db suffix from database off. If all this is feature and not bug, then it should definitely be mentioned somewhere... Jarry -- ___ This mailbox accepts e-mails only from selected mailing-lists! Everything else is considered to be spam and therefore deleted.
Re: [gentoo-user] PAM auto-adding .db suffix: feature or bug?
On 23/08/2014 21:21, Jarry wrote: Hi Gentoo-users, after playing with PAM for a few days I discovered one strange thing: PAM adds .db suffix to database definition on its own! What I mean is following: _ more /etc/pam.d/postfix authrequired pam_userdb.so db=/etc/postfix/virtual_users.db account required pam_userdb.so db=/etc/postfix/virtual_users.db ls -l /etc/postfix/virtual_users.db -rw--- 1 root root 12288 Aug 20 18:45 /etc/postfix/virtual_users.db testsaslauthd -u user1 -p password1 -s postfix 0: NO authentication failed tail -n1 saslauth.log saslauthd: pam_userdb(postfix:auth): user_lookup: could not open database '/etc/postfix/virtual_users.db': no such file or directory __ Then I changed /etc/pam.d/postfix to: authrequired pam_userdb.so db=/etc/postfix/virtual_users account required pam_userdb.so db=/etc/postfix/virtual_users And now everything works. But there is no /etc/postfix/virtual_users file, only /etc/postfix/virtual_users.db! So it seems to me PAM adds .db to database-definition itself and user has to define it without .db, otherwise what PAM is actually looking for is /etc/postfix/virtual_users.db.db Is this normal? Never seen this Windows-like behaviour with auto-adding suffix in *nix world. And I did not find anything about it in PAM-documentation. It says just: pam_userdb.so db=/path/database Not a single word that user has to cut .db suffix from database off. If all this is feature and not bug, then it should definitely be mentioned somewhere... Consult Google and find out more about postfix's postmaps with particular reference to how it's done if you use db files from sleepycat -- Alan McKinnon alan.mckin...@gmail.com
Re: [gentoo-user] Pin a package to a binary (quickpkg'd) version?
On 23/08/2014 14:42, Tanstaafl wrote: On 8/23/2014 8:16 AM, Alan McKinnon alan.mckin...@gmail.com wrote: On 23/08/2014 12:34, Tanstaafl wrote: Is it possible to do this? Not directly. I'm assuming you mean packages you built yourself and quick-pkg'ed them, not something available as a -bin Correct... I have buildpkg feature enabled in make.conf, so everything gets quickpkg'd... You can use emerge -K, so emerge will fail if there's no binpkg available. This will do what you want as long as you a) always use the -K option b) don't try emerge something else as well Portage is designed to build your packages from source; binpkgs are very much a third class citizen with only very primitive levels of support. Bummer... What I want is to be able to pin a specific package to the quickpkg'd version, so it doesn't get updated during an emerge world... Short version: You can't Longer version: You can if you package mask every version of everything greater than what you quickpkg'ed. This is tedious beyond belief Alan's crafty hack version: If you can guarantee that every package to be installed has a suitable (and correct wrt USE) quickpkg, then use emerge -K always on the machines to install the quickpkg -- Alan McKinnon alan.mckin...@gmail.com
Re: [gentoo-user] PAM auto-adding .db suffix: feature or bug?
On Sat, 2014-08-23 at 21:21 +0200, Jarry wrote: Hi Gentoo-users, after playing with PAM for a few days I discovered one strange thing: PAM adds .db suffix to database definition on its own! What I mean is following: _ more /etc/pam.d/postfix authrequired pam_userdb.so db=/etc/postfix/virtual_users.db account required pam_userdb.so db=/etc/postfix/virtual_users.db ls -l /etc/postfix/virtual_users.db -rw--- 1 root root 12288 Aug 20 18:45 /etc/postfix/virtual_users.db testsaslauthd -u user1 -p password1 -s postfix 0: NO authentication failed tail -n1 saslauth.log saslauthd: pam_userdb(postfix:auth): user_lookup: could not open database '/etc/postfix/virtual_users.db': no such file or directory __ Then I changed /etc/pam.d/postfix to: authrequired pam_userdb.so db=/etc/postfix/virtual_users account required pam_userdb.so db=/etc/postfix/virtual_users And now everything works. But there is no /etc/postfix/virtual_users file, only /etc/postfix/virtual_users.db! So it seems to me PAM adds .db to database-definition itself and user has to define it without .db, otherwise what PAM is actually looking for is /etc/postfix/virtual_users.db.db Is this normal? Never seen this Windows-like behaviour with auto-adding suffix in *nix world. And I did not find anything about it in PAM-documentation. It says just: pam_userdb.so db=/path/database Not a single word that user has to cut .db suffix from database off. If all this is feature and not bug, then it should definitely be mentioned somewhere... Jarry I think it's doing it correctly, because postfix doesn't lookup text files. Instead it expects hash files which are to be generated using stuff like newaliases (for /etc/mail/aliases.db). -- Nilesh Govindrajan m...@nileshgr.com
[gentoo-user] blocker between openrc and procps
hi, How can I resolve this blocker: (sys-process/procps-3.3.9::gentoo, installed) pulled in by sys-process/procps required by (dev-db/mysql-5.5.39::gentoo, installed) sys-process/procps required by @system (sys-apps/openrc-0.13.1::gentoo, ebuild scheduled for merge) pulled in by sys-apps/openrc required by @system sys-apps/openrc required by @selected sys-apps/openrc required by (virtual/service-manager-0::gentoo, installed) =sys-apps/openrc-0.12 required by (net-misc/netifrc-0.2.2::gentoo, ebuild scheduled for merge) Thank you very much in advance for any help! Best regards, mcc
Re: [gentoo-user] PAM auto-adding .db suffix: feature or bug?
On 24-Aug-14 0:07, Alan McKinnon wrote: On 23/08/2014 21:21, Jarry wrote: Hi Gentoo-users, after playing with PAM for a few days I discovered one strange thing: PAM adds .db suffix to database definition on its own! What I mean is following: _ more /etc/pam.d/postfix authrequired pam_userdb.so db=/etc/postfix/virtual_users.db account required pam_userdb.so db=/etc/postfix/virtual_users.db ls -l /etc/postfix/virtual_users.db -rw--- 1 root root 12288 Aug 20 18:45 /etc/postfix/virtual_users.db testsaslauthd -u user1 -p password1 -s postfix 0: NO authentication failed tail -n1 saslauth.log saslauthd: pam_userdb(postfix:auth): user_lookup: could not open database '/etc/postfix/virtual_users.db': no such file or directory __ Then I changed /etc/pam.d/postfix to: authrequired pam_userdb.so db=/etc/postfix/virtual_users account required pam_userdb.so db=/etc/postfix/virtual_users And now everything works. But there is no /etc/postfix/virtual_users file, only /etc/postfix/virtual_users.db! So it seems to me PAM adds .db to database-definition itself and user has to define it without .db, otherwise what PAM is actually looking for is /etc/postfix/virtual_users.db.db Is this normal? Never seen this Windows-like behaviour with auto-adding suffix in *nix world. And I did not find anything about it in PAM-documentation. It says just: pam_userdb.so db=/path/database Not a single word that user has to cut .db suffix from database off. If all this is feature and not bug, then it should definitely be mentioned somewhere... Consult Google and find out more about postfix's postmaps with particular reference to how it's done if you use db files from sleepycat This has nothing to do with postfix, I used it just as an example. I had the very same problem with ftp. My Postfix uses cyrus-sasl for smtp-authentification, and cyrus-sasl in my example uses pam with user-db. Problem is PAM configuration, which must contain: db=/etc/postfix/virtual_users (without .db) But real user database is in: /etc/postfix/virtual_users.db This I find very strange... Jarry -- ___ This mailbox accepts e-mails only from selected mailing-lists! Everything else is considered to be spam and therefore deleted.
