NetBSD-Bugs archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

kern/40088: uvm_fault while using tmpfs



>Number:         40088
>Category:       kern
>Synopsis:       uvm_fault while using tmpfs
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Dec 03 00:55:00 +0000 2008
>Originator:     Chris Gilbert
>Release:        NetBSD -current 2008-12-02
>Organization:
>Environment:
NetBSD 5.99.4 NetBSD 5.99.4 (GENERIC) #3: Tue Dec  2 21:32:04 GMT 2008  
root@:/usr/obj/sys/arch/amd64/compile/GENERIC amd64

>Description:
While building on a 24 core machine (having just disabled 12 cores) tmpfs 
generated a uvm_fault:
uvm_fault(0xffff80012b442e70, 0x0, 1) -> e
fatal page fault in supervisor mode
trap type 6 code 0 rip ffffffff804471b7 cs 8 rflags 10286 cr2  10 cpl 7 rsp 
ffff80012a5f3840

db{4}> bt
turnstile_block() at netbsd:turnstile_block+0x147
mutex_vector_enter() at netbsd:mutex_vector_enter+0x318
tmpfs_alloc_vp() at netbsd:tmpfs_alloc_vp+0x29
tmpfs_lookup() at netbsd:tmpfs_lookup+0x17d
VOP_LOOKUP() at netbsd:VOP_LOOKUP+0x63
getcwd_common() at netbsd:getcwd_common+0x304
sys___getcwd() at netbsd:sys___getcwd+0xc3
syscall() at netbsd:syscall+0xb6

I mounted tmpfs as:
mount -t tmpfs tmpfs /tmpfs

The server has 64GB of RAM.  At the time it tmpfs had a full source and objdir 
on it.

The server also has no Swap configured, so no dump was made.

No ECC errors are shown in the iLO/IML logs.
>How-To-Repeat:
build and delete multiple builds.

I'd probably done ~6 builds on the server before the panic happened.

Note that I'd ctrl^C'd the currently running build, which had just started.

>Fix:



Home | Main Index | Thread Index | Old Index