2.4.20: Broken AGP initialization for i845G chipset [patch]

2.4.20: Broken AGP initialization for i845G chipset [patch]

[Michael Milligan]

Chipset detection for the new Intel i845G chipset was added into the AGPGART
driver, but it appears to call the wrong initialization routine.  Stock 
compile
causes some bad interaction that messes up the PCI bus such that, in my 
case,
the follow on initialization of the Adaptec SCSI card (PCI) hangs the 
box.  Needless
to say, that's very bad since that's where all the hard drives live. 
SCSI driver
initialization happens right after the AGP initialization... from dmesg 
(after my patch):
(without the patch, it would hang before "scsi0: Adaptec ..." or before 
finding
  any drives)
....
Linux video capture interface: v1.00
Linux agpgart interface v0.99 (c) Jeff Hartmann
agpgart: Maximum main memory to use for agp memory: 203M
agpgart: Detected Intel i845G chipset
agpgart: AGP aperture is 256M @ 0xe0000000
SCSI subsystem driver Revision: 1.00
PCI: Found IRQ 9 for device 02:0a.0
PCI: Sharing IRQ 9 with 02:0a.1
PCI: Found IRQ 9 for device 02:0a.1
PCI: Sharing IRQ 9 with 02:0a.0
scsi0 : Adaptec AIC7XXX EISA/VLB/PCI SCSI HBA DRIVER, Rev 6.2.8
         <Adaptec 3950B Ultra2 SCSI adapter>
         aic7896/97: Ultra2 Wide Channel A, SCSI Id=7, 32/253 SCBs

scsi1 : Adaptec AIC7XXX EISA/VLB/PCI SCSI HBA DRIVER, Rev 6.2.8
         <Adaptec 3950B Ultra2 SCSI adapter>
         aic7896/97: Ultra2 Wide Channel B, SCSI Id=7, 32/253 SCBs

   Vendor: IBM       Model: DNES-309170Y      Rev: SAH0
   Type:   Direct-Access                      ANSI SCSI revision: 03
(scsi0:A:0): 80.000MB/s transfers (40.000MHz, offset 30, 16bit)
....

Patch below.  Calls the 845 initialization function instead of the 830MP,
and a small formatting cleanup.  This is verified working.

$ diff -u kernel-source-2.4.20/drivers/char/agp/agpgart_be.c.org 
kernel-source-2.4.20/drivers/char/agp/agpgart_be.c
--- kernel-source-2.4.20/drivers/char/agp/agpgart_be.c.org 2002-11-28 
16:53:12.000000000 -0700
+++ kernel-source-2.4.20/drivers/char/agp/agpgart_be.c  2002-12-08 
20:39:57.000000000 -0700
@@ -4551,12 +4551,12 @@
                 "Intel",
                 "i830M",
                 intel_830mp_setup },
-    { PCI_DEVICE_ID_INTEL_845_G_0,
+       { PCI_DEVICE_ID_INTEL_845_G_0,
                  PCI_VENDOR_ID_INTEL,
                  INTEL_I845_G,
                  "Intel",
                  "i845G",
-                intel_830mp_setup },
+                intel_845_setup },
         { PCI_DEVICE_ID_INTEL_840_0,
                 PCI_VENDOR_ID_INTEL,
                 INTEL_I840,

Please note... kernel version below is that of my development box (where 
kernel was compiled)

-- System Information
Debian Release: testing/unstable
Kernel Version: Linux shadow 2.4.19 #3 SMP Mon Nov 25 20:48:02 MST 2002 
i686 Pentium III (Coppermine) GenuineIntel GNU/Linux

Versions of the packages kernel-source-2.4.20 depends on:
ii  binutils       2.13.90.0.10-1 The GNU assembler, linker and binary 
utiliti
ii  bzip2          1.0.2-1        A high-quality block-sorting file 
compressor
ii  coreutils      4.5.2-1        The GNU core utilities
ii  fileutils      4.5.2-1        GNU file management utilities


Regards,
Mike

-- 
Michael Milligan  --  Free Agent  --  milli@acmeps.com

Re: 2.4.20: Broken AGP initialization for i845G chipset [patch]

[Alan Cox]

On Thu, 2002-12-19 at 23:38, Michael Milligan wrote:
> Chipset detection for the new Intel i845G chipset was added into the AGPGART
> driver, but it appears to call the wrong initialization routine.  Stock 
> compile

Ok I sent one of those to Marcelo for 2.4.21pre2. I'm not sure if I got
both

Re: 2.4.20: Broken AGP initialization for i845G chipset [patch]

[Dave Jones]

On Thu, Dec 19, 2002 at 04:38:00PM -0700, Michael Milligan wrote:
 > 
 > Patch below.  Calls the 845 initialization function instead of the 830MP,
 > and a small formatting cleanup.  This is verified working.

With testgart/some other AGP using app ?
 
It looks totally logical. I'm just wondering if it was a cut-n-paste
accident, or someone had a genuine reason for doing that in the
first place.

		Dave

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs

Re: 2.4.20: Broken AGP initialization for i845G chipset [patch]

[Michael Milligan]

Dave Jones wrote:
> On Thu, Dec 19, 2002 at 04:38:00PM -0700, Michael Milligan wrote:
>  > 
>  > Patch below.  Calls the 845 initialization function instead of the 830MP,
>  > and a small formatting cleanup.  This is verified working.
> 
> With testgart/some other AGP using app ?

Patched, It boots on my shiny new ASUS P4B533-VM motherboard with 
GeForce 4 TI 4200 AGP card.  Identifies the chipset properly and 
proceeds to boot fine.  X works (nvidia driver), OpenGL works.  Without 
patch, it hangs (the PCI bus I think) after AGP initialization.

> It looks totally logical. I'm just wondering if it was a cut-n-paste
> accident, or someone had a genuine reason for doing that in the
> first place.

I have no idea if the 830MP initialization is supposedly more compatible 
with the 845G chipset than the 845 initialization.  I just know it 
didn't work and took a guess as to what it should be.

Regards,
Mike

-- 
Michael Milligan  --  Free Agent  --  milli@acmeps.com