Patch and comments in next post (is there a way to upload a file when posting a new bug?)
Created attachment 91057 [details, diff] patch against ati-drivers-8.26.18-r1 in tree as of 20060706 - fixes bug #139450 / bug #139152 (they are dupes) - fixes bug #139456 - misc fixes: --+ depends on x11-apps/xauth (or virtual/x11) only if acpi USE flag is on --+ only installs authatieventsd.sh if acpi USE flag in on (the script which needs xauth)
err, that's bug #139452 not bug #139152
atieventsd and initscript are now installed as wished. Have you seen http://bugs.gentoo.org/show_bug.cgi?id=139436 ? Then the script also assumes that low power state is 1 and high power state 3, but that is different for different cards (mine has high power state=2). Perhaps mention it in the comments?
Created attachment 91071 [details, diff] same as above plus patches for bug #139436 - same as above plus patches for bug #139436 - sed script to change path of aticonfig to /opt/ati/bin/aticonfig
comment #3 - the second patch posted here should fix bug #139436 -- but it does not yet address different power states. Considering it is a file in etc, maybe we should, as you mentioned, post some einfo telling people to configure it for their cards.
I get two errors: > QA Notice: USE Flag 'acpid' not in IUSE for x11-drivers/ati-drivers-8.26.18-r1 shouldn't that be acpi? >install: cannot stat `atieventsd': No such file or directory
comment #6: > I get two errors: > QA Notice: USE Flag 'acpid' not in IUSE for x11-drivers/ati-drivers-8.26.18-r1 shouldn't that be acpi? ... are you using the second patch posted here against the tree or in an overlay? if it's in an overlay, did you add the overlay in make.conf? If it's against the tree, did you inadvertently emerge --sync after patching, thus wiping away the patched version? >install: cannot stat `atieventsd': No such file or directory not sure what would've caused this -- answering the above will help figure it out.
comment #6: also, check bug #139462 for a patch to fix the init.d script
Created attachment 91205 [details, diff] Fixes the ati-powermode powerstate restore issue This patch should be able to restore the high power state, regardless of it's number (2, 3, etc.).
Shouldn't the ebuild in the tree be updated with all these fixes? What can be done for to make this happen?
Yes, it should. I'm going to guess that there is some lack of caring going on, maybe, because the ati-drivers aren't free software. I don't know ... lu_zero is the maintainer for the ati-drivers ebuild, maybe he could chime in. Personally, I would like to see at least version 8.26.18-r1 stabilize at some point -- it's the first version that works reliably with my x1600.
Created attachment 93289 [details, diff] Patch against ati-drivers-8.27.10.ebuild currently in tree as of 20060802 - x11-apps/xauth is only needed if 'acpi' USE flag is set; fixed RDEPEND to reflect this - X11_IMPLEM_V was never being set, so the wrong version of the driver was installed; fixed this; ***IF YOUR X SERVER REFUSES TO LOAD AND TERMINATES COMPLAINING ABOUT THE MODULE, THIS IS WHY*** - removed old comment lines for patches needed for an older version - removed old sed that fixed ati-powermode.sh, and replaced it with a patch that implmented the fix (from Christian Affolter, comment #9) - everything inside the ./common/etc directory should not be copied out; fix to explicitly copy and install specific files - cp and donitd actions better served with newinitd; fixed
Created attachment 93290 [details] ati-drivers-8.27.10-r1.ebuild - Full-file ebuild, which has patch above (for those who have trouble patching) Full-file ebuild, which has patch above (for those who have trouble patching)
Created attachment 93291 [details] Fixes for completely broken /etc/init.d/atieventd currently in tree Copy this to OVERLAY/x11-drivers/ati-drivers/files/atieventsd.rc6, and use with above ebuild.
Note that my new changes, for the "updated" 8.27.10 ebuild fix bugs #139450, #139456, #139462, and this bug #139458. I'm raising the severity since the current ebuild in the tree will render xorg 7.0 installations unoperable. Also, changed the bug name, to be much more specific.
Thanks, I've added this to cvs (hope you don't mind Luca).
*** Bug 139450 has been marked as a duplicate of this bug. ***
*** Bug 139456 has been marked as a duplicate of this bug. ***
*** Bug 139462 has been marked as a duplicate of this bug. ***
Created attachment 93429 [details] Local output The -r1 doesn't work, as was the (no -r) one... this is a very crappy bug by the way.
http://bugs.gentoo.org/show_bug.cgi?id=139436 this is similar to the above bug, No one has yest noticed that the script calls "finger" which is not installed by default.
comment #20 - looks possibly like your kernel is "too new"; give it a downgrade and try it out.
In response to comment #21 ... chrb: can you please commit the last patch at bug #139436 ? I've posted one there that includes the previous fixes found at that bug, plus some other power state fixes, and using logger instead of echo. ... you are already half way to being my favorite person in the world for getting the fixed up ebuild in the tree. :-)