Date: Wed, 21 Apr 1999 10:26:48 -0700 From: Ted Faber <faber@ISI.EDU> To: Ben Black <black@layer8.net> Cc: freebsd-hackers@freebsd.org, paul@originative.co.uk (Paul Richards) Subject: Re: PCnet/PCI still broken in -stable? Message-ID: <199904211726.KAA15016@boreas.isi.edu> In-Reply-To: Your message of "Wed, 21 Apr 1999 04:49:21 PDT." <19990421044920.J98028@layer8.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Ben Black wrote:
>I've got an IBM 325 server with integrated PCnet/PCI ethernet which
>worked fine under 3.1-release, but now does nothing but spew
>device timeout -- resetting messages in -stable. I saw an exchange from
>September indicating a patch to fix a similar sounding problem.
>
>Any ideas?
Sure. Try this. I submitted a similar patch to Paul Richards, and
although it works for him in general, someone's having a problem with
a combination SCSI-PCnet/PCI ethernet so he hasn't committed it.
Paul: this patch is slightly different than the one I mailed you
originally. For some reason, unfathomable to me now, the patch I sent
you has the driver becoming a bus master *after* it calls the pci
attach routine, not before. This is almost certainly a bug, and I
wouldn't be surprised if it was causing the problems with the SCSI
card (of course, I wouldn't be surprised if it wasn't :-)). Give this
new patch a try and see if it helps.
The patch is relative to the /sys directory on 3.1-RELEASE, but you
should be able to apply it manually if it doesn't work on -stable.
Let me know of any problems or comments.
*** pci/if_lnc_p.c.orig Mon Mar 8 21:38:19 1999
--- pci/if_lnc_p.c Wed Apr 21 10:13:39 1999
***************
*** 72,88 ****
int unit;
{
unsigned iobase;
void *lnc; /* device specific data for interrupt handler ... */
- /* pci_map_port correctly initializes bridge chips -- tvf */
-
if ( !pci_map_port(config_id,PCI_MAP_REG_START,(u_short *)&iobase) )
printf("lnc%d: pci_port_map_attach failed?!\n",unit);
lnc = lnc_attach_ne2100_pci(unit, iobase);
if (!lnc)
return;
-
if(!(pci_map_int(config_id, lncintr_sc, (void *)lnc, &net_imask))) {
free (lnc, M_DEVBUF);
return;
--- 72,95 ----
int unit;
{
unsigned iobase;
+ unsigned data; /* scratch to make this device a bus master*/
void *lnc; /* device specific data for interrupt handler ... */
if ( !pci_map_port(config_id,PCI_MAP_REG_START,(u_short *)&iobase) )
printf("lnc%d: pci_port_map_attach failed?!\n",unit);
+
+ /* Make this device a bus master. This was implictly done by
+ pci_map_port under 2.2.x -- tvf */
+
+ data = pci_cfgread(config_id, PCIR_COMMAND, 4);
+ data |= PCIM_CMD_PORTEN | PCIM_CMD_BUSMASTEREN;
+ pci_cfgwrite(config_id, PCIR_COMMAND, data, 4);
+
lnc = lnc_attach_ne2100_pci(unit, iobase);
+
if (!lnc)
return;
if(!(pci_map_int(config_id, lncintr_sc, (void *)lnc, &net_imask))) {
free (lnc, M_DEVBUF);
return;
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199904211726.KAA15016>