Bug 34433 - [SNB bisected]Fail to type characters in xterm on a HuronRiver(0116 rev 09)
Summary: [SNB bisected]Fail to type characters in xterm on a HuronRiver(0116 rev 09)
Status: CLOSED DUPLICATE of bug 33487
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/Intel (show other bugs)
Version: unspecified
Hardware: All Linux (All)
: high normal
Assignee: Jesse Barnes
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2011-02-18 02:56 UTC by meng
Modified: 2017-10-06 14:52 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments
dmesg (36.08 KB, text/plain)
2011-02-18 02:56 UTC, meng
no flags Details
flush blt ring and set notify bit (1.31 KB, patch)
2011-02-22 14:29 UTC, Jesse Barnes
no flags Details | Splinter Review
fix flush_dw usage (1.40 KB, patch)
2011-02-22 14:39 UTC, Jesse Barnes
no flags Details | Splinter Review
add missing SNB FBC bits (1.77 KB, patch)
2011-02-23 10:31 UTC, Jesse Barnes
no flags Details | Splinter Review

Description meng 2011-02-18 02:56:28 UTC
Created attachment 43514 [details]
dmesg 

System Environment:
--------------------------
Platform:          HuronRiver(only)
Libdrm:		  (master)2.4.23-8-g36d4939343d8789d9066f7245fa2d4fe69119dd8
Mesa:		  (master)f55d027ac2e0423eba5d0664cc36668520597703
Xf86_video_intel: (master)2.14.0-26-g23f9b14df7c102c1036134835dd5d1a508059858
Cairo:		  (master)4d4056872db94573183473610ad1d81d5439fdc6
Kernel:	          (drm-intel-next) 9035a97a32836d0e456ddafaaf249a844e6e4b5e

Bug detailed description:
-------------------------
When typing characters in xterm on a HuronRiver(0116 rev 09), it is not
displayed.If moving the mouse,the characters that we has just typed will appear.
The issue doesn't exist in gnome-desktop.Pls see attached dmesg.
Specifically:
1.It works fine on Sugarbay and other HuronRiver(0126 rev08).
2.It's kernel regression.9c04f015ebc2cc2cca5a4a576deb82a311578edc is the first bad commit.

commit 9c04f015ebc2cc2cca5a4a576deb82a311578edc
Author: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Date:   Wed Dec 15 15:42:32 2010 +0800

    drm/i915: Add frame buffer compression on Sandybridge
    Add frame buffer compression on Sandybridge. The method is similar to
    Ironlake, except that two new registers of type GTTMMADR must be written
    with the right fence info.
    Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>


Reproduce steps:
-------------------------
1.xinit&
Comment 1 Chris Wilson 2011-02-18 04:44:13 UTC
Are you absolutely certain, as we fixed this once already:

commit 4efe070896e1f7373c98a13713e659d1f5dee52a
Author: Jesse Barnes <jbarnes@virtuousgeek.org>
Date:   Tue Jan 18 11:25:41 2011 -0800

    drm/i915: make the blitter report buffer modifications to the FBC unit
    
    Without this change, blits to the front buffer won't invalidate FBC
    state, causing us to scan out stale data.  Make sure we update these
    bits on every FBC enable, since they may get clobbered if we shut off
    the display.
    
    References: https://bugzilla.kernel.org/show_bug.cgi?id=26932
    Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

and I don't see the effect here.
Comment 2 meng 2011-02-20 20:53:51 UTC
Test with Kernel: (drm-intel-next)4efe070896e1f7373c98a13713e659d1f5dee52a,the 
problem exists too.
Comment 3 Jesse Barnes 2011-02-22 12:03:47 UTC
What are the values of the FBC registers when this failure occurs?  Also register 0x42020 may contain clues.
Comment 4 Jesse Barnes 2011-02-22 12:12:45 UTC
Also, Chris, since this is xterm, are we hitting sw fallbacks and going through a CPU fence register for these writes?  Or will we still use the blit engine to upload them into the front buffer on this path (as we would for gnome-terminal, which presumably goes through an XCopyArea or similar).
Comment 5 Jesse Barnes 2011-02-22 14:29:49 UTC
Created attachment 43681 [details] [review]
flush blt ring and set notify bit

Can you try this patch?
Comment 6 Jesse Barnes 2011-02-22 14:39:22 UTC
Created attachment 43682 [details] [review]
fix flush_dw usage

Oops, try this one instead, I was missing MI_FLUSH_DW args in the last one.
Comment 7 meng 2011-02-22 21:34:04 UTC
(In reply to comment #6)
> Created an attachment (id=43682) [details]
> fix flush_dw usage
> Oops, try this one instead, I was missing MI_FLUSH_DW args in the last one.

Test in (drm-intel-next)f4166442e7533ca2668d639d939b1bbdb8e6b423 with this patch,the problem still exists.Register 0x42020:0x100000A0,and by cat i915_fbc_status:FBC enabled.
Comment 8 Chris Wilson 2011-02-23 01:18:58 UTC
(In reply to comment #4)
> Also, Chris, since this is xterm, are we hitting sw fallbacks and going through
> a CPU fence register for these writes?  Or will we still use the blit engine to
> upload them into the front buffer on this path (as we would for gnome-terminal,
> which presumably goes through an XCopyArea or similar).

Yeah, xterm may either be using PolyImageGlyphs or RenderGlyphs depending upon the font and user configuration. And so it may be accessing the frontbuffer via either direct CPU writes or (effectively) CopyArea BLTs.
Comment 9 Jesse Barnes 2011-02-23 10:31:42 UTC
Created attachment 43723 [details] [review]
add missing SNB FBC bits

Ok, still missing some bits.  Can you try this one?  It adds a couple of chicken bits to 0x42020 that are supposed to be needed.
Comment 10 meng 2011-02-24 00:55:56 UTC
(In reply to comment #9)
> Created an attachment (id=43723) [details]
> add missing SNB FBC bits
> Ok, still missing some bits.  Can you try this one?  It adds a couple of
> chicken bits to 0x42020 that are supposed to be needed.

Test in commit 710f957 with this patch(id=43723),the problem still exists.And register 0x42020:0x100003A0.
Comment 11 Chris Wilson 2011-02-24 09:35:16 UTC

*** This bug has been marked as a duplicate of bug 33487 ***
Comment 12 Elizabeth 2017-10-06 14:52:55 UTC
Closing old verified.


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.