VirtualBox

Opened 13 years ago

Closed 13 years ago

#8809 closed defect (worksforme)

SATA AHCI broken in windows XP in 4.0.6

Reported by: cschmidt Owned by:
Component: virtual disk Version: VirtualBox 4.0.6
Keywords: Cc:
Guest type: Windows Host type: Linux

Description

After upgrading from VirtualBox 4.0.4 to 4.0.6 a Windows XP VM is unable to access a secondary drive (pops up the error message "S:\ is not accessible The Semaphore timeout period has expired"). The SATA driver used is Intel 82801HEM/HBM version 7.0.0.1020 from 12.0.2.2007.

A second disk seems to be affected worse than the primary, since I didn't encounter the Semaphore timeouts on C:, just on the second drive. However, accessing the c: drive also has latencies of minutes when e.g. starting a cmd window.

Attachments (1)

VBox.log.1 (58.9 KB ) - added by cschmidt 13 years ago.
Logfile for the problematic VM

Download all attachments as: .zip

Change History (7)

by cschmidt, 13 years ago

Attachment: VBox.log.1 added

Logfile for the problematic VM

comment:1 by Frank Mehnert, 13 years ago

Component: othervirtual disk

comment:2 by Frank Mehnert, 13 years ago

Are you 100% sure that this bug is related to upgrading from VBox 4.0.4 to VBox 4.0.6? I can't see any changes in the AHCI emulation which would explain this behavior.

comment:3 by cschmidt, 13 years ago

Yes, after downgrading back to 4.0.4 the timeouts were gone. At the same time, there were no noticeable performance issues with linux guests using AHCI. Just that Windows XP.

in reply to:  3 comment:4 by Frank Mehnert, 13 years ago

Replying to cschmidt:

Yes, after downgrading back to 4.0.4 the timeouts were gone. At the same time, there were no noticeable performance issues with linux guests using AHCI. Just that Windows XP.

Would you mind trying again with VBox 4.0.6 to rule out any coincidence?

comment:5 by cschmidt, 13 years ago

I'm sorry, I completely migrated away from virtualbox. I encountered a severe memory leak in this VM as well (700-800MB/day) which caused a kernel panic on the host once the VM hit 4GB, or every 2-3 days. This was completely impossible to sustain in a production environment.

comment:6 by Frank Mehnert, 13 years ago

Resolution: worksforme
Status: newclosed

Already tried to reproduce this memory leak (copying files on shared folders, right?) but without success. Without a testcase we can't do much. Actually 4.0.6 fixed a memory leak.

Note: See TracTickets for help on using tickets.

© 2023 Oracle
ContactPrivacy policyTerms of Use