spinlock problem

5 messages, 2 authors, 2008-09-17 · open the first message on its own page

spinlock problem

From: Caner Altinbasak <hidden>
Date: 2008-09-17 11:44:11

Hi!
I am using a Curtiss Wright DY4-183 board with PPC7447 CPU. I've a pci
graphics card with a M9 GPU on it. There is a transparent bridge between the
GPU and my PCI bus. My problem a trouble with radeon_dri driver. I want to
run a miniglx application to see how the graphics performance is but I am
having the following error. I think there is a problem with unhandled
interrupts! Have an idea?

-bash-3.2# ./sample_server &
[1] 1082
-bash-3.2# [miniglx] probed chipset 0x4c66
got MMIOAddress 0x1801e000 offset 67108864
shared virtual width is 1024
[drm] added 8192 byte SAREA at 0x61036000
[drm] mapped SAREA 0x61036000 to 0x18026000, size 8192
[drm] framebuffer handle = 0x80000000
[drm] register handle = 0x70000000
[pci] 8192 kB allocated with handle 0x62187000
[pci] ring handle = 0x62187000
[pci] ring read ptr handle = 0x62288000
[pci] vertex/indirect buffers handle = 0x62289000
[pci] GART texture map handle = 0x62489000
Using 8 MB AGP aperture
Using 1 MB for the ring buffhStart = 1040, hEnd = 1136, hTotal = 1312
vStart = 769, vEnd = 772, vTotal = 800
h_total_disp = 0x7f00a3    hsync_strt_wid = 0x8c040d
v_total_disp = 0x2ff031f           vsync_strt_wid = 0x830300
pixclock = 12699
freq = 7874
freq = 7874, PLL min = 12000, PLL max = 35000
ref_div = 5, ref_clk = 2700, output_freq = 15748
ref_div = 5, ref_clk = 2700, output_freq = 15748
post div = 0x1
fb_div = 0x1d
ppll_div_3 = 0x1001d
er
Using 2 MB for BUG: soft lockup detected on CPU#0!
NIP: 40042044 LR: 40042140 CTR: 400185B4
REGS: 5e25bbb0 TRAP: 0901   Not tainted  (2.6.16.13.CWCEC.K)
MSR: 00009032 <EE,ME,IR,DR>  CR: 2490C142  XER: 20000000

34 more lines

Re: spinlock problem

From: Marco Stornelli <hidden>
Date: 2008-09-17 11:55:29

But does the test work? Maybe the kernel is configured with softlook
check (see the kernel menu under kernel hacking), try to disable it and
recompile. However there's something strange...

Caner Altinbasak ha scritto:

Hi!
I am using a Curtiss Wright DY4-183 board with PPC7447 CPU. I've a pci
graphics card with a M9 GPU on it. There is a transparent bridge between the
GPU and my PCI bus. My problem a trouble with radeon_dri driver. I want to
run a miniglx application to see how the graphics performance is but I am
having the following error. I think there is a problem with unhandled
interrupts! Have an idea?

-bash-3.2# ./sample_server &
[1] 1082
-bash-3.2# [miniglx] probed chipset 0x4c66
got MMIOAddress 0x1801e000 offset 67108864
shared virtual width is 1024
[drm] added 8192 byte SAREA at 0x61036000
[drm] mapped SAREA 0x61036000 to 0x18026000, size 8192
[drm] framebuffer handle = 0x80000000
[drm] register handle = 0x70000000
[pci] 8192 kB allocated with handle 0x62187000
[pci] ring handle = 0x62187000
[pci] ring read ptr handle = 0x62288000
[pci] vertex/indirect buffers handle = 0x62289000
[pci] GART texture map handle = 0x62489000
Using 8 MB AGP aperture
Using 1 MB for the ring buffhStart = 1040, hEnd = 1136, hTotal = 1312
vStart = 769, vEnd = 772, vTotal = 800
h_total_disp = 0x7f00a3    hsync_strt_wid = 0x8c040d
v_total_disp = 0x2ff031f           vsync_strt_wid = 0x830300
pixclock = 12699
freq = 7874
freq = 7874, PLL min = 12000, PLL max = 35000
ref_div = 5, ref_clk = 2700, output_freq = 15748
ref_div = 5, ref_clk = 2700, output_freq = 15748
post div = 0x1
fb_div = 0x1d
ppll_div_3 = 0x1001d
er
Using 2 MB for BUG: soft lockup detected on CPU#0!
NIP: 40042044 LR: 40042140 CTR: 400185B4
REGS: 5e25bbb0 TRAP: 0901   Not tainted  (2.6.16.13.CWCEC.K)
MSR: 00009032 <EE,ME,IR,DR>  CR: 2490C142  XER: 20000000

43 more quoted lines

-- 
Marco Stornelli
Embedded Software Engineer
CoRiTeL - Consorzio di Ricerca sulle Telecomunicazioni
http://www.coritel.it

marco.stornelli@coritel.it
+39 06 72582838

Re: spinlock problem

From: Marco Stornelli <hidden>
Date: 2008-09-17 11:57:57

Oops, I meant soft lockup...

Marco Stornelli ha scritto:

But does the test work? Maybe the kernel is configured with softlook
check (see the kernel menu under kernel hacking), try to disable it and
recompile. However there's something strange...

Caner Altinbasak ha scritto:

quoted

Hi!
I am using a Curtiss Wright DY4-183 board with PPC7447 CPU. I've a pci
graphics card with a M9 GPU on it. There is a transparent bridge between the
GPU and my PCI bus. My problem a trouble with radeon_dri driver. I want to
run a miniglx application to see how the graphics performance is but I am
having the following error. I think there is a problem with unhandled
interrupts! Have an idea?

-bash-3.2# ./sample_server &
[1] 1082
-bash-3.2# [miniglx] probed chipset 0x4c66
got MMIOAddress 0x1801e000 offset 67108864
shared virtual width is 1024
[drm] added 8192 byte SAREA at 0x61036000
[drm] mapped SAREA 0x61036000 to 0x18026000, size 8192
[drm] framebuffer handle = 0x80000000
[drm] register handle = 0x70000000
[pci] 8192 kB allocated with handle 0x62187000
[pci] ring handle = 0x62187000
[pci] ring read ptr handle = 0x62288000
[pci] vertex/indirect buffers handle = 0x62289000
[pci] GART texture map handle = 0x62489000
Using 8 MB AGP aperture
Using 1 MB for the ring buffhStart = 1040, hEnd = 1136, hTotal = 1312
vStart = 769, vEnd = 772, vTotal = 800
h_total_disp = 0x7f00a3    hsync_strt_wid = 0x8c040d
v_total_disp = 0x2ff031f           vsync_strt_wid = 0x830300
pixclock = 12699
freq = 7874
freq = 7874, PLL min = 12000, PLL max = 35000
ref_div = 5, ref_clk = 2700, output_freq = 15748
ref_div = 5, ref_clk = 2700, output_freq = 15748
post div = 0x1
fb_div = 0x1d
ppll_div_3 = 0x1001d

49 more quoted lines

-- 
Marco Stornelli
Embedded Software Engineer
CoRiTeL - Consorzio di Ricerca sulle Telecomunicazioni
http://www.coritel.it

marco.stornelli@coritel.it
+39 06 72582838

Re: spinlock problem

From: Caner Altinbasak <hidden>
Date: 2008-09-17 12:26:58

If I do not compile with soft lockup debug option checked, I do not see the
debug messages but system freezes.

It is the same in both case, it just freezes after these messages. Soft lock
debug option only shows me call trace, instruction dump when soft lock up
happens.

There is a strange thing in call trace. I tried to follow the code, but call
trace is meaningless, most of those functions are not calling the one in
next stack. It seems just like an recursive function creating a buffer
overflow. But I think, the main reason is, handle_IRQ_event function does
not handle the IRQ, so it just enters the IRQ over and over? Any ideas? Am I
wrong? Can you suggest me a path to follow?


2008/9/17 Marco Stornelli [off-list ref]

Oops, I meant soft lockup...

Marco Stornelli ha scritto:

quoted

But does the test work? Maybe the kernel is configured with softlook
check (see the kernel menu under kernel hacking), try to disable it and
recompile. However there's something strange...

Caner Altinbasak ha scritto:

quoted

Hi!
I am using a Curtiss Wright DY4-183 board with PPC7447 CPU. I've a pci
graphics card with a M9 GPU on it. There is a transparent bridge between

the

quoted

GPU and my PCI bus. My problem a trouble with radeon_dri driver. I want

to

quoted

run a miniglx application to see how the graphics performance is but I

am

quoted

having the following error. I think there is a problem with unhandled
interrupts! Have an idea?

-bash-3.2# ./sample_server &
[1] 1082
-bash-3.2# [miniglx] probed chipset 0x4c66
got MMIOAddress 0x1801e000 offset 67108864
shared virtual width is 1024
[drm] added 8192 byte SAREA at 0x61036000
[drm] mapped SAREA 0x61036000 to 0x18026000, size 8192
[drm] framebuffer handle = 0x80000000
[drm] register handle = 0x70000000
[pci] 8192 kB allocated with handle 0x62187000
[pci] ring handle = 0x62187000
[pci] ring read ptr handle = 0x62288000
[pci] vertex/indirect buffers handle = 0x62289000
[pci] GART texture map handle = 0x62489000
Using 8 MB AGP aperture
Using 1 MB for the ring buffhStart = 1040, hEnd = 1136, hTotal = 1312
vStart = 769, vEnd = 772, vTotal = 800
h_total_disp = 0x7f00a3    hsync_strt_wid = 0x8c040d
v_total_disp = 0x2ff031f           vsync_strt_wid = 0x830300
pixclock = 12699
freq = 7874

66 more quoted lines

Re: spinlock problem

From: Marco Stornelli <hidden>
Date: 2008-09-17 13:26:08


Caner Altinbasak ha scritto:

If I do not compile with soft lockup debug option checked, I do not see the
debug messages but system freezes.

It is the same in both case, it just freezes after these messages. Soft lock
debug option only shows me call trace, instruction dump when soft lock up
happens.

There is a strange thing in call trace. I tried to follow the code, but call
trace is meaningless, most of those functions are not calling the one in
next stack. It seems just like an recursive function creating a buffer
overflow.

I think they are the interrupt nested paths.

But I think, the main reason is, handle_IRQ_event function does

not handle the IRQ, so it just enters the IRQ over and over? Any ideas? Am I
wrong? Can you suggest me a path to follow?

What about the function cw_pic_unmask_irq_183? Have you looked at it?

2008/9/17 Marco Stornelli [off-list ref]

quoted

Oops, I meant soft lockup...

Marco Stornelli ha scritto:

quoted

But does the test work? Maybe the kernel is configured with softlook
check (see the kernel menu under kernel hacking), try to disable it and
recompile. However there's something strange...

Caner Altinbasak ha scritto:

quoted

Hi!
I am using a Curtiss Wright DY4-183 board with PPC7447 CPU. I've a pci
graphics card with a M9 GPU on it. There is a transparent bridge between

the

quoted

GPU and my PCI bus. My problem a trouble with radeon_dri driver. I want

to

quoted

run a miniglx application to see how the graphics performance is but I

am

quoted

having the following error. I think there is a problem with unhandled
interrupts! Have an idea?

-bash-3.2# ./sample_server &
[1] 1082
-bash-3.2# [miniglx] probed chipset 0x4c66
got MMIOAddress 0x1801e000 offset 67108864
shared virtual width is 1024
[drm] added 8192 byte SAREA at 0x61036000
[drm] mapped SAREA 0x61036000 to 0x18026000, size 8192
[drm] framebuffer handle = 0x80000000
[drm] register handle = 0x70000000
[pci] 8192 kB allocated with handle 0x62187000
[pci] ring handle = 0x62187000
[pci] ring read ptr handle = 0x62288000
[pci] vertex/indirect buffers handle = 0x62289000
[pci] GART texture map handle = 0x62489000
Using 8 MB AGP aperture
Using 1 MB for the ring buffhStart = 1040, hEnd = 1136, hTotal = 1312
vStart = 769, vEnd = 772, vTotal = 800
h_total_disp = 0x7f00a3    hsync_strt_wid = 0x8c040d

68 more quoted lines

-- 
Marco Stornelli
Embedded Software Engineer
CoRiTeL - Consorzio di Ricerca sulle Telecomunicazioni
http://www.coritel.it

marco.stornelli@coritel.it
+39 06 72582838

`h`	back out one level
`j`	next message in thread
`k`	previous message in thread
`l`	drill in
`Esc`	close help / fold thread tree
`?`	toggle this help