On Mon, Dec 02, 2013 at 04:55:38PM +0100, Robert Millan wrote: > On 02/12/2013 16:04, Steven Chamberlain wrote: > > On 02/12/13 14:55, Robert Millan wrote: > >> Perhaps we need to replace that patch with the dummy "return 0" version? > >> I'm afraid the check might fail if we don't, since upstream doesn't use > >> it at all. > > > > Ahhh probably, this could be why AMD Radeon didn't work for me, it > > failed to detect KMS, but I was using libdrm2 from sid with the > > preliminary patch. > > Hi Kostik, > > We've noticed that your patch for drmCheckModesettingSupported(): > > http://people.freebsd.org/~kib/drm/libdrm.2.patch > > is not currently being used, as upstream doesn't have it, and the > FreeBSD port uses a simpler, "return 0" stub instead: Whay do you mean by upstream ? The freedesktop libdrm git ? > > http://svnweb.freebsd.org/ports/head/graphics/libdrm/files/extra-xf86drmMode.c?revision=300896&view=markup > > Is this patch still meant to be used with current FreeBSD? > > We're not even sure if the patch works anymore. If the "return 0" stub > is the way to go, we could push it upstream and also apply it in Debian > to follow what FreeBSD is doing. Again, what do you mean by 'anymore' ? It worked when I wrote it, and kernel driver(s) definitely were not changed there. OTOH, I am not sure if radeon driver for drm2 followed the protocol, it should call drm_add_busid_modesetting() for the detection to work. Apparently, I do not see the corresponding call from the radeon driver in HEAD. This should be fixed. Also, I have no idea why ports do not use the suggested patch, and did the gross hack with 'return (0);'. I even not sure whom to ask.
Attachment:
pgpPI2wNSy_4p.pgp
Description: PGP signature