The attached patchset drops support for pre-mach64 and non-pci cards from xf86-video-ati. Is it based on the work by Luc Verhaegen, but focuses on code deletions: 4 files changed, 3 insertions(+), 87 deletions(-) 5 files changed, 23 insertions(+), 0 deletions(-) 4 files changed, 368 insertions(+), 662 deletions(-) <- sane probe 1 files changed, 1 insertions(+), 222 deletions(-) 3 files changed, 26 insertions(+), 35 deletions(-) 9 files changed, 24 insertions(+), 1051 deletions(-) 7 files changed, 41 insertions(+), 285 deletions(-) 12 files changed, 26 insertions(+), 378 deletions(-) 21 files changed, 33 insertions(+), 258 deletions(-) 9 files changed, 12 insertions(+), 184 deletions(-) 38 files changed, 61 insertions(+), 3281 deletions(-) <- pci mach64 only 2 files changed, 2 insertions(+), 342 deletions(-) I think it's decent, except for a (ridiculously) big commit, trying to split that results in partially deleting code that soon gets deleted in whole. Also, it has been tested only on an ATI Rage Pro. Please document what you think.
Created attachment 8205 [details] keep pci mach64 only tgz patchset
Is this that version you once showed me where you took my code, dropped Mach64 CX and GX support and merged it with recent changes?
(In reply to comment #2) > Is this that version you once showed me where you took my code, dropped Mach64 > CX and GX support and merged it with recent changes? yes, it's the same version with the CX and GX support back in.
Oh, i see. Heh. History repeating (no, not the CX/GX stuff).
Created attachment 8227 [details] keep pci mach64 only - try 1 split big commit a little, not it is: 38 files changed, 42 insertions(+), 2121 deletions(-)
Created attachment 8228 [details] keep pci mach64 only - try 1 split big commit a little, now it is: 38 files changed, 42 insertions(+), 2121 deletions(-)
Created attachment 8229 [details] keep pci mach64 only - try 2 split the big commit once again, but this time it is along the files touched rather than functionality: 15 files changed, 26 insertions(+), 1395 deletions(-)
Created attachment 8250 [details] keep pci mach64 only - try 3 cosmetic.
Created attachment 8322 [details] keep pci mach64 only - try 4 drop a superfluous function.
Created attachment 8400 [details] keep pci mach64 only - try 5 rebase to master.
Created attachment 8460 [details] keep pci mach64 only - try 6 minor cosmetic
The master branch is open for development, and you're the de facto atimisc maintainer...
Comment on attachment 8460 [details] keep pci mach64 only - try 6 This is now tracked in the mach64-pci branch of xf86-video-ati: http://gitweb.freedesktop.org/?p=xorg/driver/xf86-video-ati.git;a=shortlog;h=ma ch64-pci
I guess the branch has to receive some testing on some other card/architecure before merging to master as it has been only tested on RagePro/i386. Also, it probably needs some kind of consent from distro people. My line here goes like: the AVOID_NON_PCI configure variable is always set since X.org 7.0 because of a typo in configure.ac and mach32 is not accelerated, so the only effective functionality loss should be mach32 video modes. Lastly, there is a quirk in that sparse I/O conflict avoidance was dropped and replaced with the "probe_sparse" option.
This is now tracked in: http://gitweb.freedesktop.org/?p=xorg/driver/xf86-video-ati.git;a=shortlog;h=mach64-pci-1 Put all the grep-based commits first.
Merged to master (9d77aabdff919360f0c9333105436c31f1f5749a).
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.