Re: pci_request_regions() failure

[Alec H. Peterson]

--On Thursday, May 6, 2004 3:50 PM -0400 "Richard B. Johnson" 
<root@xxxxxxxxxxxxxxxxxx> wrote:

> The BIOS should have aligned everything correctly.
> 0xec108fff - 0xec107000 = 0x1fff (0x2000 bytes)
>
> 0xec107000 / 0x2000 = 0x76083, * 0x2000 = 0xec106000
> (where it should have been). Check to see if that region is
> clear and if it is, write that address to the PCI
> BAR. If it isn't, check the next higher address
> (ex106000 + 0x2000), etc.

That region of memory is actually pretty tightly packed, but I think I have 
managed to solve the problem.

In drivers/pcmcia/yenta.c, it seems that the regions of memory in question 
are also too small in yenta_allocate_res() (BRIDGE_SIZE_MIN) in addition to 
not being properly aligned.  So, I added the following clause to check for 
both cases (since yenta_allocate_res() will find a new block if necessary):


       start = config_readl(socket, offset) & mask;
       end = config_readl(socket, offset+4) | ~mask;
#if 1
       if (!(type & IORESOURCE_IO) && (((end - start) < BRIDGE_SIZE_MIN) ||
           (start & (end - start))))
       {
               printk(KERN_INFO "yenta %s: Preassigned resource start %lx 
end %lx too small or not aligned.\n", socket->dev->slot_name, start, end);
               res->start = res->end = 0;
       }
       else
#endif
       if (start && end > start) {
               res->start = start;

Now everything is working perfectly, at least for me.  I realize that this 
sort of thing might break other stuff, but it seems to me that this sort of 
check is a good idea regardless.

Note that this is my first attempt at a kernel patch, so please let me know 
if this is horribly wrong or otherwise just a bad idea.

Thanks!

Alec

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