2.3.18: offset in pci structure. Am I the only one?

Carlo E. Prelz (fluido@fluidware.com)
Sun, 12 Sep 1999 19:52:34 +0300


Good day. After skipping releases 2.3.1[567], for lack of time (needed
to upgrade to Yet Another Packet Filtering System), today I at last
decided to give a try to 2.3.18. but I experienced a strange behavior,
that is repeated with plain 2.3.18, with 2.3.18ac2, and with Martin's
pci patches applied.

At boot, both my pci ne2k card and matrox millennium II cards are not
correctly detected. The second one gives the following message:

> matroxfb: cannot ioremap(C049, 16384), matroxfb disabled

while the net2k driver returns:

> No useable cards found, driver NOT installed.

Since it was more urgent for me to have the net card working, I
recompiled both things as modules, and started experiment with the
ne2k module. I found out that the call to check_region was not
satisfied, so I added a printk to find out what address was
returned. Well, after the call to pci_find_class
pdev->resource[0].start contained 0xd400c048, and pdev->irq contained
0xc0000. The real values, as found in /proc/pci, are 0xd400 and 0xc
(12).

It appears that the pci_dev structure is somehow 2-byte offset. After
finding that I believed I'd find some similar cases on the kernel
list, but I could not find references to similar accidents. I looked
at pci_find_class, but it just forwards a pointer from a linked
list. The next step would be to find where this pci device list is
filled up. Sadly, time's over for today, so I thought I'd send a
message to the list. Maybe the solution is easy and/or well known. In
the mean time, I am back to the safe haven of 2.3.14, and will most
probably remain here until the next weekend... But I'd like to find
out if I am really alone out here!

Carlo

-- 
  *         Se la Strada e la sua Virtu' non fossero state messe da parte,
* K * Carlo E. Prelz - fluido@fluidware.com         che bisogno ci sarebbe
  *               di parlare tanto di amore e di rettitudine? (Chuang-Tzu)

- 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/