On Mon, May 27, 2013 at 10:25:21AM +0800, Fam Zheng wrote: > On Fri, 05/24 09:07, Richard W.M. Jones wrote: > > On Fri, May 24, 2013 at 01:36:55PM +0800, Fam Zheng wrote: > > > Changes from v5: > > > 05: Rename bs to s for BDRVCURLState. > > > 06: Use int64_t for offsets. > > > Fix printf format string. > > > Move introducing of use_count to 07. > > > 07: Drop explicit cast. > > > Use int64_t for offsets. > > > Use_count moved here. > > > 08: Remove duplicated. > > > Move s->url = NULL to separate patch. > > > 09: Fix while(0); > > > 12: Added: > > > curl: set s->url to NULL after free. > > > > This patch is definitely not working. The guest sees loads of > > disk errors like this: > > > > [ 30.880265] sd 2:0:0:0: [sda] > > [ 30.880265] Add. Sense: I/O process terminated > > [ 30.880265] sd 2:0:0:0: [sda] CDB: > > [ 30.880265] Read(10): 28 00 00 bf b0 87 00 00 01 00 > > [ 32.030702] sd 2:0:0:0: [sda] > > [ 32.031663] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE > > [ 32.031663] sd 2:0:0:0: [sda] > > [ 32.031663] Sense Key : Aborted Command [current] > > [ 32.031663] sd 2:0:0:0: [sda] > > [ 32.031663] Add. Sense: I/O process terminated > > [ 32.031663] sd 2:0:0:0: [sda] CDB: > > [ 32.031663] Read(10): 28 00 00 00 08 46 00 00 01 00 > > [ 32.031663] blk_update_request: 32 callbacks suppressed > > [ 32.031663] end_request: I/O error, dev sda, sector 2118 > > [ 32.031663] quiet_error: 46 callbacks suppressed > > [ 32.031663] Buffer I/O error on device sda1, logical block 2055 > > [ 32.031663] sd 2:0:0:0: [sda] > > [ 32.031663] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE > > [ 32.031663] sd 2:0:0:0: [sda] > > [ 32.031663] Sense Key : Aborted Command [current] > > [ 32.031663] sd 2:0:0:0: [sda] > > [ 32.031663] Add. Sense: I/O process terminated > > [ 32.031663] sd 2:0:0:0: [sda] CDB: > > [ 32.031663] Read(10): 28 00 00 00 08 45 00 00 01 00 > > [ 32.031663] end_request: I/O error, dev sda, sector 2117 > > [ 32.031663] Buffer I/O error on device sda1, logical block 2054 > > [ 32.031663] sd 2:0:0:0: [sda] > > > > Rich, are you testing with libguestfs or qemu? Since I can't get > libguestfs run with your patch to refuse writable open.
You need to add the --ro option and make sure LIBGUESTFS_BACKEND=direct. > Can you post your command? Sure, I'm using the attached test script. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming blog: http://rwmj.wordpress.com Fedora now supports 80 OCaml packages (the OPEN alternative to F#)
test.sh
Description: Bourne shell script
