Re: XIO: fatal IO error 35 (Resource temporarily unavailable) on X server ":0.0"

On Sat, Mar 26, 2011 at 07:29:32PM +0100, Hartmann, O. wrote:
> On 03/26/11 19:19, Gary Jennejohn wrote:
> >On Sat, 26 Mar 2011 18:53:52 +0100
> >"O. Hartmann"<ohartman_at_mail.zedat.fu-berlin.de>  wrote:
> >
> >>On 03/26/11 14:48, Kostik Belousov wrote:
> >>>On Sat, Mar 26, 2011 at 01:44:52PM +0100, O. Hartmann wrote:
> >>>>On 26.03.2011 11:55, Andriy Gapon wrote:
> >>>>>on 26/03/2011 11:10 O. Hartmann said the following:
> >>>>>>Updating ports and source of my FreeBSD 9.0-CURRENT/amd64 host, 
> >>>>>>equipted
> >>>>>>with a
> >>>>>>AMD/ATi HD4830 graphics board driven by X11 xf86-video-ati driver 
> >>>>>>(which
> >>>>>>has
> >>>>>>been recently update as far as I saw),
> >>>>>>resulted this morning in an 'un-login-able' box.
> >>>>>>
> >>>>>>I see the xdm-login requester, but after successfully login, I see 
> >>>>>>for a
> >>>>>>second
> >>>>>>the desktop (windowmaker), but X11 immediately dies and resets to the 
> >>>>>>xdm
> >>>>>>requester again.
> >>>>>Are you subscribed to our x11 list?  It's quite low volume.
> >>>>>Please see recent messages in its archive, perhaps they could help you.
> >>>>>
> >>>>
> >>>>Hello.
> >>>>
> >>>>No, I'm actually not subscribed to that list. Since the problem occured
> >>>>since yesterday's update out of the blue, I thought it could be more
> >>>>FreeBSD related than X11 related. But I will reply to the X11 list as
> >>>>well. I simply forgot in my desperate pain, sorry.
> >>>>
> >>>See
> >>>http://lists.freebsd.org/pipermail/freebsd-x11/2011-March/010737.html
> >>
> >>I downgraded the AMD/ATi driver to version xf86-video-ati-6.14.0_1, I
> >>desperately need my boxes since I have to finish work.
> >>Three different types of AMD/ATi graphics cards out of the same product
> >>class are not working with the new xf86-video-ati-6.14.1 and as the
> >>above mentioned link suggests, the solution is well known. I was
> >>wondering how this port could be so easily slipped into the repository ...
> >>
> >>Since the patch Jun-uk Kim referred to is a GIT patch and it seems to
> >>work, is there a 'simple' way without installing the GIT stuff from
> >>ports to patch the defective port?
> >>
> >Have you looked at the link?  The patch is trivial and can easily be
> >applied by hand.
> >
> Yes I did.

Put the following content into the ports/x11-drivers/xf86-video-ati/files/patch-all


From fe2e0ad3ffa58f40311319c950b842e2928a5740 Mon Sep 17 00:00:00 2001
From: matthew green <mrg_at_eterna.com.au>
Date: Mon, 21 Mar 2011 16:17:58 +0000
Subject: bug fix for r6xx/r7xx UMS

Signed-off-by: Alex Deucher <alexdeucher_at_gmail.com>
---
diff --git src/r600_exa.c src/r600_exa.c
index 0ed0ced..7736d24 100644
--- src/r600_exa.c
+++ src/r600_exa.c
_at__at_ -621,8 +621,12 _at__at_ R600PrepareCopy(PixmapPtr pSrc,   PixmapPtr pDst,
 	return FALSE;
 
     if (accel_state->same_surface == TRUE) {
+#if defined(XF86DRM_MODE)
 	unsigned height = RADEON_ALIGN(pDst->drawable.height,
 				       drmmode_get_height_align(pScrn, accel_state->dst_obj.tiling_flags));
+#else
+	unsigned height = pDst->drawable.height;
+#endif
 	unsigned long size = height * accel_state->dst_obj.pitch * pDst->drawable.bitsPerPixel/8;
 
 #if defined(XF86DRM_MODE)
diff --git src/radeon_textured_video.c src/radeon_textured_video.c
index c886ed0..d247db6 100644
--- src/radeon_textured_video.c
+++ src/radeon_textured_video.c
_at__at_ -248,7 +248,11 _at__at_ RADEONPutImageTextured(ScrnInfoPtr pScrn,
     BoxRec dstBox;
     int dst_width = width, dst_height = height;
     int aligned_height;
+#ifdef XF86DRM_MODE
     int h_align = drmmode_get_height_align(pScrn, 0);
+#else
+    int h_align = 1;
+#endif
     /* make the compiler happy */
     s2offset = s3offset = srcPitch2 = 0;
 
--
cgit v0.8.3-6-g21f6