AMD PCSCSI (AM53c974) driver bug in 2.2.x in SMP but not

David A. Ranch (dranch@trinnet.net)
Fri, 09 Apr 1999 16:36:45 -0700


Hello kernel gurus..

I hope this post gets on the kernel list since I'm not subscribed.
Because of this, please CC: any replies to dranch@trinnet.net

--

Anyway..

I have confirmed a problem with the AMD PC-SCSI driver hanging in the 2.2.x kernels when compiled in SMP mode but -NOT- in non-SMP mode. It should also be noted that a 2.0.36 kernel with SMP support works fine.

This is very reproducable on every boot.

--
Hardware:

- Hp Vectra XU 5/90C - Dual Pentium 90 (hehehe) + 48MB RAM + Imagine 128 PCI video + integrated AMD PCscsi controller + integrated AMD PCnet32 controller + newest BIOS installed

- Linux 2.2.5ac6 (tried main 2.2.5 and other 2.2.5 AC patches)

- Redhat 5.2 with all the recent patches

Error:

In 2.2.5AC6 kernel with SMP enabled, the box initially boots, sees both processors, network, ide, blah blah, etc and just after it says:

"Freeing unused kernel memory: 44k freed"

The SCSI LED locks solid and I periodically get:

-------------------------------------------------------------- SCSI host 0 abort (PID 44) timed out - resetting SCSI bus is being reset for Host 0 Channel 0 AM53C974_RESET CALLED AM53C974 registry dump:

IO BASE: 0x1ff00 CTCREG: 0x120000 CMDREG: 0x42 STATREG: 0x00 ISREF:0xC0 CFIREG: 0x07 CNTLREG1: 0x57 CNTLREG2: 0x40 CNTLREG3: 0x18 CNTLREG4: 0x44 DMACMP: 0xC3 DMASTC: 0x0400; DMASPA 0x019C00 DMAWBC: 0x000 DMAWAC: 0x19000 DMASTATUS: 0x00 --------------------------------------------------------------

Any thoughts? This seems like it is a SMP specific issue. If needed, I can try a recommended 2.1.x kernel and since this is a play box, I can help test new patches, etc.

Thanks for the help!

--David .----------------------------------------------------------------------------. | David A. Ranch - Linux/Networking/PC hardware dranch@trinnet.net | !---- ----! `----- For more detailed info, see http://www.ecst.csuchico.edu/~dranch -----'

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/