Hello,
On Mon, 14 Dec 1998, Michele Comitini wrote:
> Hi,
>
> I have tried a 2.1.126 (old but it was the only one I had handy) and
> it worked as a charm!
The patch below will hopefully make 2.0.3x work as well for you:
--- /usr/src/linux-2.0.36/arch/i386/kernel/bios32.c Sun Oct 19 18:47:21 1997
+++ ./bios32.c Sun Dec 20 10:24:06 1998
@@ -871,6 +871,15 @@
int i, length;
/*
+ * We should always prefer direct methods against BIOS based ones.
+ * So, if any direct method is available, we want to go with it
+ * and just ignore the PCI BIOS.
+ */
+ access_pci = check_direct_pci();
+ if (access_pci)
+ goto end;
+
+ /*
* Follow the standard procedure for locating the BIOS32 Service
* directory by scanning the permissible address range from
* 0xe0000 through 0xfffff for a valid BIOS32 structure.
@@ -910,5 +919,6 @@
if (bios32_entry && check_pcibios())
access_pci = &pci_bios_access;
#endif
+end:
return memory_start;
}
-------------------------- Cut Here --------------------------
> On 12-Dec-98 Gerard Roudier wrote:
>
> > The 876 is a multi-function PCI device (2 functions). There are 2
> > independant devices in the Silicium that share the same PCI slot.
> > The driver, in fact, just sees 2 independant devices that shall be
> > reported by the PCI code as 2 PCI device functions.
> > If the driver only detects 1 device, then some PCI code may well be
> > buggy somewhere.
>
> Yes it must be the BIOS' PCI code, now I see that the kernel
Using the PCI BIOS services is useless with common x86 chipsets.
So we should not even want to hear about it to be bogus or not on
this point.
> finds 2 function on this 876. The mb is brand new ASUS super socket 7 with
> Ali chipset. I will try to see if there is a new BIOS on the Asus site,
> and try that too and let you know.
[ ... ]
> >> Anyway this new driver seems very good and fast! Thanks Roudier!
> >
> > Only fast ? :-)
>
> Ok, it is the FASTEST! I have better performance on a P200MMX with
> sym875 than on a 533 Alpha with AHA-2940, same disks, same kernels 2.0.36.
I am interested in the result.
Alpha (at least earlier implementations) is not able to do memory access
smaller than 32 bits and required some sparce space to be used for 16 bits
and 8 bits MMIO access. This makes use of MMIO more complex and, AFAIK,
PCI device drivers use normal IO under Alpha. This may explain at least
part of the difference.
> I will probably swap the two controllers when time permits.
Adaptec controllers BIOS will complain if the CHS mapping is not among the
ones defined by Adaptec. The Linux driver will choose a CHS mapping based
on the disk capacity and some EEPROM flag. You have to be careful about
that.
> The solution sym8xx + your drivers is very good and cost effective.
It is the main reason I have much fun working on these drivers.
Regards,
Gerard.
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to [EMAIL PROTECTED]
