On 02/07/2012 03:47 PM, Stefan Hajnoczi wrote:
On Wed, Feb 01, 2012 at 08:37:12AM +0530, Supriya Kannery wrote:
+ /* stash state before reopen */
+ raw_rs->stash_s = g_malloc0(sizeof(BDRVRawState));
+ memcpy(raw_rs->stash_s, s, sizeof(BDRVRawState));
Copying a struct is fragile, Mike Roth pointed out the potential issue
with aligned_buf.
If raw-posix could open from a given file descriptor as an alternative
to opening a filename, then it would be clean and natural to simply
re-initialize from the dup'd file descriptor in the abort case. That's
the approach I would try instead of stashing the whole struct.
Stefan
fine, will get V1 with stashing of only required fields wherever
possible instead of stashing the full struct.
