Bug 91921

Summary: radeon: command stream overflowed with current git
Product: Mesa Reporter: octoploid <octoploid>
Component: Drivers/Gallium/r600Assignee: Default DRI bug account <dri-devel>
Status: RESOLVED FIXED QA Contact: Default DRI bug account <dri-devel>
Severity: normal    
Priority: medium CC: notasas, octoploid
Version: git   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:
Attachments: screenshot
candidate fix

Description octoploid 2015-09-08 16:47:18 UTC
Created attachment 118141 [details]
screenshot

Trying to run krunner (from KDE plasma-workspace-5.4.1) results in 
a garbled dialog box (see attached screenshot).

Console shows:

radeon: command stream overflowed
radeon: The kernel rejected CS see dmesg for more information.

dmesg shows nothing.

[    2.109692] [drm] Initialized drm 1.1.0 20060810
[    2.109703] [drm] radeon kernel modesetting enabled.
[    2.109903] [drm] initializing kernel modesetting (RS780 0x1002:0x9614 0x1043:0x834D).
[    2.109907] [drm] register mmio base: 0xFBEE0000
[    2.109907] [drm] register mmio size: 65536
[    2.110434] ATOM BIOS: 113
[    2.110453] radeon 0000:01:05.0: VRAM: 128M 0x00000000C0000000 - 0x00000000C7FFFFFF (128M used)
[    2.110455] radeon 0000:01:05.0: GTT: 512M 0x00000000A0000000 - 0x00000000BFFFFFFF
[    2.110457] [drm] Detected VRAM RAM=128M, BAR=128M
[    2.110458] [drm] RAM width 32bits DDR
[    2.110523] [TTM] Zone  kernel: Available graphics memory: 4081244 kiB
[    2.110524] [TTM] Zone   dma32: Available graphics memory: 2097152 kiB
[    2.110525] [TTM] Initializing pool allocator
[    2.110528] [TTM] Initializing DMA pool allocator
[    2.110544] [drm] radeon: 128M of VRAM memory ready
[    2.110545] [drm] radeon: 512M of GTT memory ready.
[    2.110552] [drm] Loading RS780 Microcode
[    2.110557] == power state 0 ==
[    2.110558]  ui class: none
[    2.110559]  internal class: boot 
[    2.110560]  caps: video 
[    2.110561]  uvd    vclk: 0 dclk: 0
[    2.110561]          power level 0    sclk: 50000 vddc_index: 2
[    2.110562]          power level 1    sclk: 50000 vddc_index: 2
[    2.110562]  status: c r b 
[    2.110563] == power state 1 ==
[    2.110564]  ui class: performance
[    2.110564]  internal class: none
[    2.110565]  caps: video 
[    2.110565]  uvd    vclk: 0 dclk: 0
[    2.110566]          power level 0    sclk: 50000 vddc_index: 1
[    2.110566]          power level 1    sclk: 70000 vddc_index: 2
[    2.110566]  status: 
[    2.110567] == power state 2 ==
[    2.110567]  ui class: none
[    2.110568]  internal class: uvd 
[    2.110568]  caps: video 
[    2.110569]  uvd    vclk: 53300 dclk: 40000
[    2.110570]          power level 0    sclk: 50000 vddc_index: 1
[    2.110570]          power level 1    sclk: 50000 vddc_index: 1
[    2.110570]  status: 
[    2.123272] [drm] radeon: dpm initialized
[    2.123276] [drm] GART: num cpu pages 131072, num gpu pages 131072
[    2.141465] [drm] PCIE GART of 512M enabled (table at 0x00000000C0040000).
[    2.141512] radeon 0000:01:05.0: WB enabled
[    2.141514] radeon 0000:01:05.0: fence driver on ring 0 use gpu addr 0x00000000a0000c00 and cpu addr 0xffff880216386c00
[    2.141515] [drm] Supports vblank timestamp caching Rev 2 (21.10.2013).
[    2.141516] [drm] Driver supports precise vblank timestamp query.
[    2.141517] radeon 0000:01:05.0: radeon: MSI limited to 32-bit
[    2.141531] [drm] radeon: irq initialized.
[    2.173747] [drm] ring test on 0 succeeded in 1 usecs
[    2.174031] [drm] ib test on ring 0 succeeded in 0 usecs
[    2.174257] [drm] Radeon Display Connectors
[    2.174258] [drm] Connector 0:
[    2.174258] [drm]   VGA-1
[    2.174259] [drm]   DDC: 0x7e40 0x7e40 0x7e44 0x7e44 0x7e48 0x7e48 0x7e4c 0x7e4c
[    2.174260] [drm]   Encoders:
[    2.174260] [drm]     CRT1: INTERNAL_KLDSCP_DAC1
[    2.174261] [drm] Connector 1:
[    2.174261] [drm]   DVI-D-1
[    2.174262] [drm]   HPD3
[    2.174263] [drm]   DDC: 0x7e50 0x7e50 0x7e54 0x7e54 0x7e58 0x7e58 0x7e5c 0x7e5c
[    2.174263] [drm]   Encoders:
[    2.174263] [drm]     DFP3: INTERNAL_KLDSCP_LVTMA
[    2.174324] switching from power state:
[    2.174324]  ui class: none
[    2.174325]  internal class: boot 
[    2.174326]  caps: video 
[    2.174327]  uvd    vclk: 0 dclk: 0
[    2.174327]          power level 0    sclk: 50000 vddc_index: 2
[    2.174328]          power level 1    sclk: 50000 vddc_index: 2
[    2.174328]  status: c b 
[    2.174329] switching to power state:
[    2.174329]  ui class: performance
[    2.174330]  internal class: none
[    2.174330]  caps: video 
[    2.174331]  uvd    vclk: 0 dclk: 0
[    2.174331]          power level 0    sclk: 50000 vddc_index: 1
[    2.174332]          power level 1    sclk: 70000 vddc_index: 2
[    2.174332]  status: r 
[    2.214955] [drm] fb mappable at 0xF0141000
[    2.214955] [drm] vram apper at 0xF0000000
[    2.214956] [drm] size 7299072
[    2.214956] [drm] fb depth is 24
[    2.214957] [drm]    pitch is 6912
[    2.215011] fbcon: radeondrmfb (fb0) is primary device
Comment 1 octoploid 2015-09-08 17:23:05 UTC
Started with:

commit fbb423b43380d0e43d14056e96846da412693148
Author: Grazvydas Ignotas <notasas@gmail.com>
Date:   Thu Sep 3 01:54:28 2015 +0300

    r600g: apply disable workaround on all scissors
    
    During review of the "r600g: make all scissor states use single atom" patch
    Marek Olšák noticed that scissor disable workaround should be applied on
    all scissor states and not just first one, so let's do so.
    
    Signed-off-by: Marek Olšák <marek.olsak@amd.com>

Reverting the commit above fixes the issue.
Comment 2 Grazvydas Ignotas 2015-09-08 21:32:16 UTC
Created attachment 118156 [details] [review]
candidate fix

Does this patch help?
Comment 3 octoploid 2015-09-09 04:59:13 UTC
(In reply to Grazvydas Ignotas from comment #2)
> Created attachment 118156 [details] [review] [review]
> candidate fix
> 
> Does this patch help?

Yes. Thank you.
Comment 4 Marek Olšák 2015-09-24 17:54:13 UTC
Fixed by 8ae8feca844c953510e94f116be2fc37fb14ff. Closing.

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.