Date: Wed, 9 Apr 2003 19:45:10 -0400 From: Don Bowman <don@sandvine.com> To: "'freebsd-smp@freebsd.org'" <freebsd-smp@freebsd.org>, Don Bowman <don@sandvine.com> Subject: ichsmb SMP safe [in 4.7]? Message-ID: <FE045D4D9F7AED4CBFF1B3B813C8533701B36190@mail.sandvine.com>
next in thread | raw e-mail | index | archive | help
is the ichsmb driver SMP safe in 4.7?
I have a supermicro x5DPR-8g2+ motherboard with 2 2.8GHz XEON.
If I compile a non-SMP kernel for it, it works well on
accessing the SMB to get the Xeon temperature.
If I compile an SMP kernel for it, I get one of two things:
messages like 'ichsmb0: device timeout, status=0x42',
or else the system locks up.
It really comes down to the ichsmb_wait function i think.
I do a 'readb'. This does:
splhigh()
...
bus_space_write(...)
ichsmb_wait(...)
splx()
static int
ichsmb_wait(sc_p sc)
{
...
sleep:
error = tsleep(sc, PZERO | PCATCH, "ichsmb", hz / 3);
switch (error) {
case ERESTART:
if (sc->ich_cmd != -1)
goto sleep;
/* FALLTHROUGH */
case 0:
smb_error = sc->smb_error;
break;
case EWOULDBLOCK:
...
smb_error = SMB_ETIMEOUT;
break;
default:
smb_error = SMB_EABORT;
break;
}
return (smb_error);
}
All I can see for the hang would be the ERESTART happening?
I'm going to put a timeout in there to see what it does
for me, but am otherwise not sure why the behaviour of
SMP versus non SMP would be different for this driver?
--don
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?FE045D4D9F7AED4CBFF1B3B813C8533701B36190>