My s2895 doesn't do a soft reset correctly with the coreboot image, but it does with the factory image. Have you seen that problem? It seems strange, because the "Reset triggered" message still shows up, along with "Rebooting the emulated CPU", but then I get Readback errors. I patched SerialICE to redo the handshake on reset, but that doesn't help.
I think I'm getting closer, but if someone has already seen something similar or has suggestions...
Thanks, Myles
On 30. juni 2010 00:55, Myles Watson wrote:
My s2895 doesn't do a soft reset correctly with the coreboot image, but it does with the factory image. Have you seen that problem? It seems strange, because the "Reset triggered" message still shows up, along with "Rebooting the emulated CPU", but then I get Readback errors. I patched SerialICE to redo the handshake on reset, but that doesn't help.
I think I'm getting closer, but if someone has already seen something similar or has suggestions...
I saw something similar with the HP board I ported, turned out to be the IPMI BMC hijacking the serial port at soft reset. Not sure exactly what conditions triggered it, but I got it sorted by disabling the IPMI serial port CLI using ipmitool or something similar. Let me know if this sounds at all likely with your setup, and I'll dig some more.
On Wed, Jun 30, 2010 at 3:47 PM, Arne Georg Gleditsch arne.gleditsch@numascale.com wrote:
On 30. juni 2010 00:55, Myles Watson wrote:
My s2895 doesn't do a soft reset correctly with the coreboot image, but it does with the factory image. Have you seen that problem? It seems strange, because the "Reset triggered" message still shows up, along with "Rebooting the emulated CPU", but then I get Readback errors. I patched SerialICE to redo the handshake on reset, but that doesn't help.
I think I'm getting closer, but if someone has already seen something similar or has suggestions...
I saw something similar with the HP board I ported, turned out to be the IPMI BMC hijacking the serial port at soft reset. Not sure exactly what conditions triggered it, but I got it sorted by disabling the IPMI serial port CLI using ipmitool or something similar. Let me know if this sounds at all likely with your setup, and I'll dig some more.
I don't think my board has IPMI. It seems to be triggered by changing the HyperTransport widths. If I leave it in 8-bit mode it doesn't quit. Unfortunately that's not OK, so I'm still trying to narrow it down.
Thanks, Myles
On 06/30/2010 05:57 PM, Myles Watson wrote:
On Wed, Jun 30, 2010 at 3:47 PM, Arne Georg Gleditsch arne.gleditsch@numascale.com wrote:
On 30. juni 2010 00:55, Myles Watson wrote:
My s2895 doesn't do a soft reset correctly with the coreboot image, but it does with the factory image. Have you seen that problem? It seems strange, because the "Reset triggered" message still shows up, along with "Rebooting the emulated CPU", but then I get Readback errors. I patched SerialICE to redo the handshake on reset, but that doesn't help.
I think I'm getting closer, but if someone has already seen something similar or has suggestions...
I saw something similar with the HP board I ported, turned out to be the IPMI BMC hijacking the serial port at soft reset. Not sure exactly what conditions triggered it, but I got it sorted by disabling the IPMI serial port CLI using ipmitool or something similar. Let me know if this sounds at all likely with your setup, and I'll dig some more.
I don't think my board has IPMI. It seems to be triggered by changing the HyperTransport widths. If I leave it in 8-bit mode it doesn't quit. Unfortunately that's not OK, so I'm still trying to narrow it down.
Myles, What do you mean by quit? My 830M boards just quit/stop/halt at certian points (seems to always at same places) also with the recent versions of serialice. They used to work great with the first publicly released versions of serialice. If I cycle the power real quickly it continues on. I am not sure what changed but if this sounds like your problem too, try a earlier version. I am curious to hear your results.
From: Joseph Smith [mailto:joe@settoplinux.org] On 06/30/2010 05:57 PM, Myles Watson wrote:
On Wed, Jun 30, 2010 at 3:47 PM, Arne Georg Gleditsch arne.gleditsch@numascale.com wrote:
On 30. juni 2010 00:55, Myles Watson wrote:
My s2895 doesn't do a soft reset correctly with the coreboot image, but it does with the factory image. Have you seen that problem? It seems strange, because the "Reset triggered" message still shows up, along with "Rebooting the emulated CPU", but then I get Readback errors. I patched SerialICE to redo the handshake on reset, but that doesn't help.
I think I'm getting closer, but if someone has already seen something similar or has suggestions...
I saw something similar with the HP board I ported, turned out to be
the
IPMI BMC hijacking the serial port at soft reset. Not sure exactly
what
conditions triggered it, but I got it sorted by disabling the IPMI
serial
port CLI using ipmitool or something similar. Let me know if this
sounds at
all likely with your setup, and I'll dig some more.
I don't think my board has IPMI. It seems to be triggered by changing the HyperTransport widths. If I leave it in 8-bit mode it doesn't quit. Unfortunately that's not OK, so I'm still trying to narrow it down.
Myles, What do you mean by quit?
When it gets to a soft reset, the serial link starts having read back errors.
My 830M boards just quit/stop/halt at certian points (seems to always at same places) also with the recent versions of serialice.
Does the factory BIOS have the same behavior?
They used to work great with the first publicly released versions of serialice. If I cycle the power real quickly it continues on. I am not sure what changed but if this sounds like your problem too, try a earlier version. I am curious to hear your results.
That sounds like it could be a different problem. The factory BIOS works fine for me with the latest version of SerialICE. I tried earlier versions and had some problems.
I wonder if the filters that have gotten added are taking away some I/O that you need. Have you tried removing the parts of serialice.lua that don't apply to your board?
Thanks, Myles
-
Arne Georg Gleditsch -
Joseph Smith -
Myles Watson