Bug #42
closedkernel panic on 4way SMP system and 1.4-RELEASE
0%
Description
Hello,
Yesterday I decided to "upgrade" my Dragonfly system from the Developer
branch (2005-12-14) to the new 1.4 branch. This didn't work as I hoped.
The thing is that I don't have the time to analyse the problem (I will
rather switch back to the working version), I only post here a few
things I noticed.
The hardware is a Proliant 7000 with 4 cpu-s.
The symptoms:
- there were kernel panics, complaining about a page fault
- I saw a message: "clist_alloc_cblocks: M_NOWAIT malloc failed, trying
M_WAITOK"
- also saw this one: '[diagnostic] cache_lock: blocked on 0x3... "usr"'
'[diagnostic] cache_lock: unblocked usr'
The version I used is the 1.4 slip, cvsupped around 7th January afternoon.
Sorry, I cannot provide you more data as I need to find a workaround
ASAP, also the server is located in a weird environment where I have
little physical access, so I can't go there very frequently to reboot...
I hope the very few details I sent gave some ideas about the problem, if
so could you please share those with me?
Thank you,
David Beck.
Updated by dillon almost 19 years ago
:
:Hello,
:
:Yesterday I decided to "upgrade" my Dragonfly system from the Developer
:branch (2005-12-14) to the new 1.4 branch. This didn't work as I hoped.
:The thing is that I don't have the time to analyse the problem (I will
:rather switch back to the working version), I only post here a few
:things I noticed.
:
:The hardware is a Proliant 7000 with 4 cpu-s.
:
:The symptoms:
:
:- there were kernel panics, complaining about a page fault
:- I saw a message: "clist_alloc_cblocks: M_NOWAIT malloc failed, trying
:M_WAITOK"
:- also saw this one: '[diagnostic] cache_lock: blocked on 0x3... "usr"'
: '[diagnostic] cache_lock: unblocked usr'
:
:The version I used is the 1.4 slip, cvsupped around 7th January afternoon.
:
:Sorry, I cannot provide you more data as I need to find a workaround
:ASAP, also the server is located in a weird environment where I have
:little physical access, so I can't go there very frequently to reboot...
:
:I hope the very few details I sent gave some ideas about the problem, if
:so could you please share those with me?
:
:Thank you,
: David Beck.
:
Well, you can ignore the clist_alloc and cache_lock diagnostics, those
are harmless.
The kernel panic is another matter. We'd definitely need more
information to figure out what is going on there, such as a kernel
core and backtrace.
I would never suggest that someone do a blind update of an inaccessible
production server. Sometimes things break! Too late for you, I know,
but there isn't much I can do to help without more information. I can
only suggest that the machine be reverted until you are able to test on
a more accessible box.
-Matt
Matthew Dillon
<dillon@backplane.com>
Updated by dbeck almost 19 years ago
Matthew Dillon wrote:
I made a try with 1.5 cvsupped a few hours ago. It seems OK so far. The
reason to upgrade was that I intend to use nullfs functionality for some
jail configs. Sorry to say, I'm not going to be able to test this issue
within a few months, becuse I have only one 4way box at the moment.
Best regards,
David.