Date: Thu, 9 Dec 2021 13:23:43 +0100 From: Peter <pmc@citylink.dinoex.sub.org> To: freebsd-stable@freebsd.org Subject: [ahd driver] 12.3: kernel crash when stopping disks Message-ID: <YbH1T60u4TtzCKcJ@gate.intra.daemon.contact> In-Reply-To: <YawjnVXXXC4q%2BkV0@gate.intra.daemon.contact>
next in thread | previous in thread | raw e-mail | index | archive | help
> Dec 5 01:08:25 <local0.info> edge gstopd[64139]: Error received from stop unit command > Dec 5 01:08:25 <kern.crit> edge kernel: ahd0: Recovery Initiated - Card was not paused > Dec 5 01:08:25 <kern.crit> edge kernel: >>>>>>>>>>>>>>>>>> Dump Card State Begins <<<<<<<<<<<<<<<<< > Dec 5 01:08:25 <kern.crit> edge kernel: ahd0: Dumping Card State at program address 0x7e Mode 0x22 > Dec 5 01:08:25 <kern.crit> edge kernel: (pass0:ahd0:0:0:0): SCB 247 - timed out > Dec 5 01:08:25 <kern.crit> edge kernel: (pass0:ahd0:0:0:0): Queuing a BDR SCB > Dec 5 01:08:25 <kern.crit> edge kernel: (pass0:ahd0:0:0:0): Bus Device Reset Message Sent Hija, I had a closer look into this one: There must be a timeout flaw in the driver logic. I tried to run the STOP UNIT from camcontrol with "-t 30", but nevertheless these controller errors happen to appear after some 5 or 10 seconds. So whereever it gets the timeout from, it is not the right one. The kernel crash is then an occasional consequence of these strange timeouts - it happened only once, while the erroneous timeouts happen more often. I now workaround the issue: as the STOP UNIT is the only concerned command, I invoke that with the IMMED bit, not waiting until the disk finally stops: as this is only for saving the rain-forests (any my power bill), I don't care if or when the disks might manage to stop. No more problems or errors since that. cheerio, PMc
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?YbH1T60u4TtzCKcJ>