System panics in vol_multistepsio_read_source() due to BAD TRAP in module "vxio" due to a NULL pointer dereference
Article ID: 100005680
Updated On:
Description
Error Message
Panic string:
BAD TRAP: type=31 rp=2a100e954a0 addr=1d0 mmu_fsr=0 occurred in module "vxio" due to a NULL pointer dereference
Panic stack:
vol_multistepsio_read_source+0x98()
vol_multistepsio_start+0x550()
volkcontext_process+0x268()
vol_rv_write2_start+0x470()
voliod_iohandle+0x30()
voliod_loop+0x69c()
thread_start+4()
Cause
When paging module in VxVM gets heavily used, regions in the DCO log of the instant snapshot volume could remain invalid even after valid data is pushed from source volume.
This leads to DCO log corruption, which inturn leads to both panic and I/O failure.
Resolution
Code changes have been made to prevent DCO logs from getting corrupted. This issue is addressed via etrack incident # 2015570 / 2325547.
The fix is included in the following Veritas Volume Manager patch releases (and above):
VxVM 5.0MP3RP4HF12
VxVM 5.0MP3RP5 (tentative release date is 22nd April 2011)
Applies To
- Solaris 10 (applicable to all versions)
- VxVM volumes with DCO log
Issue/Introduction
1. System panics in vol_multistepsio_read_source() in VxVM (Veritas Volume Manager). Panic stack would look like:
vol_multistepsio_read_source+0x98()
vol_multistepsio_start+0x550()
volkcontext_process+0x268()
vol_rv_write2_start+0x470()
voliod_iohandle+0x30()
voliod_loop+0x69c()
thread_start+4()
2. I/O on specific regions of an instant snapshot volume will fail with I/O error.
Additional Information
ETrack: 2015570
Was this article helpful?
Yes
No
Powered by
