Solaris: Powerpath 4.50 & usb device causes system crash

So did Sun say the crash was due to PowerPath, running the EMCGrab or a USB device?

"MMU Miss" error is very generic and can be used to describe a variety of issues, we need more detail.

Can you post up Sun's findings and I'm sure we will be able to help?

Conor

CB

Hi Conor

I copy and paste what SUN emailed me.
Let me give you the scenario.

First I wanted to upgrade this machine to PowerPath 5.1 as my PP is on version 4.5. As a standard procedure, I ran the emcgrab which cause the server to crash.
Then SUN came in and wanted to get the crash files. SUN engineer insert in a USB thumb drive and cause the server to crash the 2nd time.

Based on the 2nd vmcore file, SUN said that it is caused by MMU miss. Let me know if you have further queries.

Below is the mail from SUN.

The document EMC pointed out is not the issue occurred in the vmcore.2 at all...

* The panic in our issue is an mmu miss and in the
bug pointed out the issue was a bad mutex issue.

*The stack and the behaviour detail in document
is completely different from what happend in vmcore.2
Looking at the stack

unix:mutex_enter+0x4()
genunix:ddi_prop_search_common+0x58(0x2000000010, 0x0, 0x1, 0xf09, 0x144db98, 0x2a101c7f408)
genunix:ddi_prop_op+0x24(0x2000000010, 0x0, 0x1, 0x9, 0x144db98, 0x2a101c7f408)
sd:sd_prop_op+0x1c4(0x2000000010, 0x0, 0x1, 0x9, 0x144db98, 0x2a101c7f408, 0x2a101c7f404)
emcp:PowerPlatformBottomDispatch+0xc38(0x30039b69468, 0x3003bd94b98, 0x20, 0x9, 0x144db98, 0x2a101c7f408)
...skipping
...in between there are only emcp functions....

emcp:power_prop_op+0x32c(0x2000000010, 0x300001db600, 0x1, 0x9, 0x144db98, 0x2a101c7f408)
genunix:cdev_prop_op+0x60(0x2000000010, 0x300001db600, 0x1, 0x9, 0x144db98, 0x2a101c7f408)
genunix:e_ddi_getprop_int64+0x44(0x2000000010, 0x3, 0x144db98, 0x9, 0xffffffffffffffff, 0x30038b8f5d8)
genunix:bdev_Size+0x18(0x2000000010, 0x1, 0x0, 0x30038b8f5d8, 0x14a4290, 0x1497400)
specfs:spec_open+0x2fc(0x2a101c7f6b8, 0x1)
pcfs:pc_getfattype+0x64(0x3003a4e48d8, , 0x2a101c7f7a0, 0x2a101c7f79c, 0x2a101c7f7a8, 0x0)
pcfs:pcfs_mount+0x6e8(0x3003b162040, 0x300390ca0b8, 0x2a101c7f9e0, 0x30038b8f5d8, 0x2a101c7f8e0, 0x0)
genunix:domount+0x738(, 0x2a101c7f9e0, 0x300390ca0b8?, 0x30038b8f5d8, 0x2a101c7f9d8)
genunix:mount+0x11c(, 0x2a101c7fae8, 0x10974, 0xff28a174, 0x5, 0xff33c000)
genunix:syscall_ap+0x6c(0xffbff62e, 0xffbff642, 0x4, 0x22264, 0x223cc, 0xc)
unix:syscall_trap32+0xa8()


the genuninx function genunix:cdev_prop_op calling emcp:power_prop_op passes to it a VALID
dev pointer....then after emcp handling the last emcp function on stack
(PowerPlatformBottomDispatch) passed to sd driver (sd:sd_prop_op) a not yet
valid dev pointer....

so EMC needs to provide a feedback on what was going on within the emcp
function involved on the stack.

Of course searching into sunsolve with any of the function of the panic
stack can let you gather tons of document ...but....
this DOES NOT mean that these documents match our panic issue and are relevant
for us.
Let me know.

Thanks Zanial,

Can you check if Solaris patch 114476-05 or a later version is installed on this host? (as per Primus solution emc118999)

Please also review emc168548 "Solaris 9 host panics with USB device attached."
This is a known issue with Solaris 9, and is resolved in Solaris 10. See SUN document 5066010 for this issue.

If either of these Solutions does not match your issue, I suggest opening a case with the software support center, they will require a grab from the host, the vmcore files, Sun's Analysis and any other relevant information.

Conor