In message: <20050608150344.GL41050_at_obiwan.tataz.chchile.org> Jeremie Le Hen <jeremie_at_le-hen.org> writes: : Hi Warner, : : On Mon, Jun 06, 2005 at 11:07:42PM -0600, M. Warner Losh wrote: : > : It seems that the ed(4) driver from 2005.01.01 are working well. : > : There have been numerous commits since then, but I'm trying to find : > : the culprit ATM. I'll keep list abreast of the results. : > : > If you find the one that's at issue, I'll be happy to work with you to : > resolve it. : : I narrowed the one breaking MAC address probe : : http://docs.freebsd.org/cgi/getmsg.cgi?fetch=1231599+0+archive/2005/cvs-all/20050220.cvs-all : : I tried to patch the driver myself and I found that you should not clear : pending interrupts before reading the MAC address or it won't work. : : I didn't made a patch as I don't know if the correct way to handle this : is to pull up the ed_nic_outb() call from ed_probe_Novell_generic() to : ed_cbus_probe() or to move the ed_Novell_read_mac() call back in the : ed_probe_Novell_generic() function. Maybe the following patch will be better than either of these alternatives? It adds the call to read the Novel NIC to the pci probe routine, clearly an oversight on the rascal that's responsible for the change that you've pointed out above. :-) Warner Index: if_ed_pci.c =================================================================== RCS file: /cache/ncvs/src/sys/dev/ed/if_ed_pci.c,v retrieving revision 1.41 diff -u -r1.41 if_ed_pci.c --- if_ed_pci.c 5 Mar 2005 18:30:10 -0000 1.41 +++ if_ed_pci.c 8 Jun 2005 16:44:24 -0000 _at__at_ -87,6 +87,7 _at__at_ ed_release_resources(dev); return (error); } + ed_Novell_read_mac(sc); error = ed_alloc_irq(dev, 0, RF_SHAREABLE); if (error) {Received on Wed Jun 08 2005 - 14:49:22 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:36 UTC