HAMMER is looking nice and stable for the release. I've moved on to
working through its error handling, getting rid of Debugger() calls
and assertions in the I/O error path, and adjusting the code so a
write error forces the mount into read-only mode.
It's kinda fun. I'm testing it by taking an external USB2 hard drive
and mounting it w/ HAMMER, then unplugging it in the middle of a big
cpdup. I've found a couple of panics in CAM related to the device
going away unexpectedly, and stuff like that, which I'm fixing as I
go.
I expect to get it working nicely by thursday afternoon and will commit
and MFC it for 2.0 then. Hopefully I'll get it to the point where
I can pop the USB connector in and out without crashing the machine
or corrupting the HAMMER filesystem. Mind you, once you pull the drive
you have to umount -f / remount HAMMER and will lose unflushed file
data, but the filesystem itself shouldn't become corrupt. I'm not
pulling the disk's power, just its USB connector :-)
-Matt
Matthew Dillon
<dillon@backplane.com>| James Bruce | Re: Dual-Licensing Linux Kernel with GPL V2 and GPL V3 |
| Peter Zijlstra | [PATCH 00/23] per device dirty throttling -v8 |
| Jan Engelhardt | intel iommu (Re: -mm merge plans for 2.6.23) |
| Peter Zijlstra | [RFC/PATCH 0/4] CPUSET driven CPU isolation |
git: | |
| Gerrit Renker | [PATCH 18/37] dccp: Support for Mandatory options |
| Rick Jones | Re: Network latency regressions from 2.6.22 to 2.6.29 |
| David Miller | [GIT]: Networking |
| Josip Rodin | bnx2_poll panicking kernel |
