Re: [9fans] Hunging rio?

Re: [9fans] Hunging rio?

[rob pike]

I would like to know which VGA is used in these systems.  I suspect
it's the driver for that particular card but in any case it would be helpful
to know if the problem is specific to a particular model.

-rob

Re: [9fans] Hunging rio?

[Andrew Pochinsky]

   From: "rob pike" <rob@plan9.bell-labs.com>
   Date: Fri, 18 May 2001 21:04:49 -0400

   I would like to know which VGA is used in these systems.  I suspect
   it's the driver for that particular card but in any case it would be helpful
   to know if the problem is specific to a particular model.

   -rob


Here is output of aux/vga -B and aux/vga -p. It's a PCI card; there is
also an i82557 NIC on the same PCI bus, if that matters.

--andrew

0xC0000 55 AA 40 EB 7B C8 30 00 00 00 00 00 00 00 00 00  U.@.{.0.........
0xC0010 00 00 00 00 00 00 00 00 54 01 00 00 00 00 49 42  ........T.....IB
0xC0020 4D 00 4D 7A 01 5F 30 08 00 00 00 00 00 00 00 00  M.Mz._0.........
0xC0030 20 37 36 31 32 39 35 35 32 30 FF FF 15 94 AE AD   761295520......
0xC0040 33 3F 00 3C 00 00 00 00 00 01 00 00 00 00 00 00  3?.<............
0xC0050 31 39 39 39 2F 31 32 2F 31 33 20 31 37 3A 35 39  1999/12/13 17:59
0xC0060 00 00 00 00 E9 F6 50 00 E9 EB 50 00 E9 AE 12 00  ......P...P.....
0xC0070 00 00 00 00 00 00 00 00 00 00 FF FF 15 94 AE AD  ................
0xC0080 E9 EA 00 0D 0A 41 54 49 20 4D 41 43 48 36 34 20  .....ATI MACH64 
0xC0090 42 49 4F 53 20 50 2F 4E 20 31 31 33 2D 34 31 39  BIOS P/N 113-419
0xC00A0 30 32 2D 31 30 37 20 0D 0A 00 28 43 29 20 31 39  02-107 ...(C) 19
0xC00B0 38 38 2D 39 37 2C 20 41 54 49 20 54 65 63 68 6E  88-97, ATI Techn
0xC00C0 6F 6C 6F 67 69 65 73 20 49 6E 63 2E 42 4B 33 2E  ologies Inc.BK3.
0xC00D0 39 2E 35 2F 33 2E 30 39 32 20 67 74 67 70 6D 63  9.5/3.092 gtgpmc
0xC00E0 2E 63 20 6D 6C 76 36 31 31 20 00 4D 41 43 48 36  .c mlv611 .MACH6
0xC00F0 34 47 50 50 43 49 4D 54 53 47 55 3F 00 00 4A 00  4GPPCIMTSGU?..J.
...
vga misc             23
vga feature          00
vga sequencer        03 01 0F 00 0A
vga crt              D5 9F AB 81 AD 13 32 52 - 00 40 00 00 00 00 00 00
                     03 26 FF A0 60 04 05 A3 - FF
vga graphics         00 00 00 00 00 50 05 0F - FF
vga attribute        00 01 02 03 04 05 06 07 - 08 09 0A 0B 0C 0D 0E 0F
                     41 FF 0F 00 00
vga vm a b           16777216        0
vga vmz               8388608
vga apz               8388608
vga linear                  1
vga->attr: 0xC00EB=MACH64GPPCIMTSGU
mach64xx pci 41650 io d800 pciregs
mach64xx ccru 300
mach64xx HTotalDisp          009F00D5
mach64xx HSyncStrtWid        001300AD
mach64xx VTotalDisp          03FF0432
mach64xx VSyncStrtWid        00060403
mach64xx VlineCrntVline      033403FF
mach64xx OffPitch            28000000
mach64xx IntCntl             80000054
mach64xx CrtcGenCntl         03010600
mach64xx OvrClr              00000000
mach64xx OvrWidLR            00000000
mach64xx OvrWidTB            00000000
mach64xx CurClr0             FFFFFFFF
mach64xx CurClr1             00000000
mach64xx CurOffset           000A0000
mach64xx CurHVposn           02C00199
mach64xx CurHVoff            00300030
mach64xx ScratchReg0         04900400
mach64xx ScratchReg1         00801000
mach64xx ClockCntl           00007C02
mach64xx BusCntl             73330100
mach64xx MemCntl             10753A7B
mach64xx ExtMemCntl          E0000C01
mach64xx MemVgaWpSel         00010000
mach64xx MemVgaRpSel         00010000
mach64xx DacRegs             00FF0000
mach64xx DacCntl             8501210A
mach64xx GenTestCntl         00000188
mach64xx ConfigCntl          00003D42
mach64xx ConfigChipId        7C004750
mach64xx ConfigStat0         00800015
mach64xx ConfigStat1         00100280
mach64xx ConfigStat2         02208000
mach64xx DspConfig           001A05ED
mach64xx DspOnOff            056005C0
mach64xx DpBkgdClr           FFFFFFFF
mach64xx DpChainMsk          FFFFFFFF
mach64xx DpFrgdClr           FFFFFFFF
mach64xx DpMix               FFFFFFFF
mach64xx DpPixWidth          FFFFFFFF
mach64xx DpSrc               FFFFFFFF
mach64xx DpWriteMsk          FFFFFFFF
mach64xx LcdIndex            00000000
mach64xx LcdData             00000000
mach64xx PLL         AD D5 1F 64 D8 03 CF DA - F5 92 DA 01 A6 1B 00 00
                     00 00 80 00 18 A2 CC 10 - 00 00 00 00 00 00 00 00
mach64xx VCLK0	25172284
mach64xx VCLK1	28289952
mach64xx VCLK2	134868018
mach64xx VCLK3	25172284
mach64xx pixel clock = 25170000
ATI BIOS rom 0x100 freq 0x0 clock 0x704
clocks: 43605 60224 51323 48 0 0 0 0 0 0 0 0 340 0 0 16969
programmable clock: 4
clock to program: 3
reference numerator: 14320
reference denominator: 1
internal clock
reference divider in plls
panelid 85 x 1 y 0

Re: [9fans] Hunging rio?

[rob pike]

Two things to try.

1) Get the system to the state when it would panic if you resized,
and then dump and save the VGA state.  The register set you sent
isn't set up yet.

2) Try running at a lower resolution.  What resolution are you at now?

-rob

Re: [9fans] Hunging rio?

[Francisco J Ballesteros]

In my case, the hang happens by resizing using top/bottom lines and
also one of the corners.
No need to reach the edge of the display to get it frozen.

However, I just changed the hw and it seems to work properly now.
I don't now if the bug may be just for a particular vga. I'll
take a look to the exact model of the hanging board and drop a line
to the list when I get back to my office. 

Ah, FTS that when it hangs I can always see the red lines drawn during
resize. that's all I can say by now. 

rob pike wrote:
> 
> I would like to understand this.  Can you characterize the error better?
> For example, does the hang occur if you don't hit the edge of the display?
> Does any resize cause a hang, or only one that touches the right or bottom
> edge?
> 
> -rob

Re: [9fans] Hunging rio?

[Andrew Pochinsky]

   It's rather time comsuming to study this error systematically, so I
can only offer some anecdots describing 'it seems that so and so
happens often'. Anyway, rio seems to hang if the first (and the only)
window is resized or created with its lrc in the lower right quadrant
of the screen. It does not need to be on the edge, just far enough to
the bottom right (It may be more than 1024 in X, but the feeling is
that one can avoid hanging rio by making the first window small enough.)

Another impression is that this bug manifests itself more often with
the very first window than with other windows. (The installation
sequence seems to be unaffected.)

As to the remark about faulty hardware -- the same machines run dec'00
distribution now without problem (without any change to hw), so, while
that proves nothing, it does not feel like a hw issue.

--andrew


   From: "rob pike" <rob@plan9.bell-labs.com>
   Date: Fri, 18 May 2001 14:22:32 -0400

   I would like to understand this.  Can you characterize the error better?
   For example, does the hang occur if you don't hit the edge of the display?
   Does any resize cause a hang, or only one that touches the right or bottom
   edge?

   -rob

Re: [9fans] Hunging rio?

[rob pike]

I would like to understand this.  Can you characterize the error better?
For example, does the hang occur if you don't hit the edge of the display?
Does any resize cause a hang, or only one that touches the right or bottom
edge?

-rob

Re: [9fans] Hunging rio?

[nemo]

[-- Attachment #1: Type: text/plain, Size: 228 bytes --]

Curious,

	I was suspecting of a hw failure, but one of my
machines also hangs some times. It's always when I keep the
side of a window dragged to resize it. It just gets frozen and
does not attend interrupts any more. 


[-- Attachment #2: Type: message/rfc822, Size: 4311 bytes --]

[-- Attachment #2.1.1: Type: text/plain, Size: 113 bytes --]

Curious, I don't see this behaviour.
Could you elaborate your running conditions a little bit more?

Kenji


[-- Attachment #2.1.2: Type: message/rfc822, Size: 2523 bytes --]

From: Andrew Pochinsky <avp@honti.mit.edu>
To: 9fans@cse.psu.edu
Subject: [9fans] Hunging rio?
Date: Thu, 17 May 2001 14:40:23 -0400 (EDT)
Message-ID: <200105171840.f4HIeNj05867@honti.mit.edu>


 I've just download the new distribution (5/17/01) and it shows the
most peculiar behaviour: once plan9 is installed, and glenda boot the
terminal, trying to resize the term window (e.g., dragging the lower
right corner to the bottom right of the screen and pausing there for a
few seconds to appreciate the size of the window before releasing
button 2) results in a completely frozen machine -- C-tC-t does not
work, only CTRL-ALT-DEL puts the end to its misery.

If rio is not started, plan 9 seems to be quite happy. Same thing
happens almost reporducibly on two machines: Thinkpad 570 and a P-4
machine with ATI video card. (In both cases vga is recognized and
everything else works just fine. If the window resize gesture is done
differently, rio sometimes passes this step and works ok after that.

Does anyone have a suggestion how to fix the problem?

--andrew

P.S. This outrage did not happen on the older distributions, if that's
     any help in locating the bug.

Re: [9fans] Hunging rio?

[okamoto]

[-- Attachment #1: Type: text/plain, Size: 113 bytes --]

Curious, I don't see this behaviour.
Could you elaborate your running conditions a little bit more?

Kenji


[-- Attachment #2: Type: message/rfc822, Size: 2504 bytes --]

From: Andrew Pochinsky <avp@honti.mit.edu>
To: 9fans@cse.psu.edu
Subject: [9fans] Hunging rio?
Date: Thu, 17 May 2001 14:40:23 -0400 (EDT)
Message-ID: <200105171840.f4HIeNj05867@honti.mit.edu>


 I've just download the new distribution (5/17/01) and it shows the
most peculiar behaviour: once plan9 is installed, and glenda boot the
terminal, trying to resize the term window (e.g., dragging the lower
right corner to the bottom right of the screen and pausing there for a
few seconds to appreciate the size of the window before releasing
button 2) results in a completely frozen machine -- C-tC-t does not
work, only CTRL-ALT-DEL puts the end to its misery.

If rio is not started, plan 9 seems to be quite happy. Same thing
happens almost reporducibly on two machines: Thinkpad 570 and a P-4
machine with ATI video card. (In both cases vga is recognized and
everything else works just fine. If the window resize gesture is done
differently, rio sometimes passes this step and works ok after that.

Does anyone have a suggestion how to fix the problem?

--andrew

P.S. This outrage did not happen on the older distributions, if that's
     any help in locating the bug.

[9fans] Hunging rio?

[Andrew Pochinsky]

 I've just download the new distribution (5/17/01) and it shows the
most peculiar behaviour: once plan9 is installed, and glenda boot the
terminal, trying to resize the term window (e.g., dragging the lower
right corner to the bottom right of the screen and pausing there for a
few seconds to appreciate the size of the window before releasing
button 2) results in a completely frozen machine -- C-tC-t does not
work, only CTRL-ALT-DEL puts the end to its misery.

If rio is not started, plan 9 seems to be quite happy. Same thing
happens almost reporducibly on two machines: Thinkpad 570 and a P-4
machine with ATI video card. (In both cases vga is recognized and
everything else works just fine. If the window resize gesture is done
differently, rio sometimes passes this step and works ok after that.

Does anyone have a suggestion how to fix the problem?

--andrew

P.S. This outrage did not happen on the older distributions, if that's
     any help in locating the bug.