Re: [libvirt] Re: [Libvir] Reopening the old discussion about virDomainBlockPeek
"Richard W.M. Jones" <rjones(a)redhat.com> wrote:
Attached is an updated patch. The changes are:
- updated to latest CVS
- run make check / syntax-check
- remove virsh subcommand (as per Dan's suggestion - see below)
- some more things that Dan pointed out - see below
...
Index: src/xend_internal.c
===================================================================
...
+ data.path = path;
+ data.ok = 0;
+
+ if (xend_parse_sexp_blockdevs (domain->conn, root,
+ priv->xendConfigVersion,
+ check_path, &data) == -1)
+ return -1;
+
+ if (!data.ok) {
+ virXendError (domain->conn, VIR_ERR_INVALID_ARG,
+ _("invalid path"));
+ return -1;
+ }
+
+ /* The path is correct, now try to open it and get its size. */
+ fd = open (path, O_RDONLY);
+ if (fd == -1 || fstat (fd, &statbuf) == -1) {
+ virXendError (domain->conn, VIR_ERR_SYSTEM_ERROR, strerror (errno));
+ goto done;
+ }
+
+ /* Seek and read. */
+ /* NB. Because we configure with AC_SYS_LARGEFILE, off_t should
+ * be 64 bits on all platforms.
+ */
+ if (lseek (fd, offset, SEEK_SET) == (off_t) -1 ||
+ saferead (fd, buffer, size) == (ssize_t) -1) {
+ virXendError (domain->conn, VIR_ERR_SYSTEM_ERROR, strerror (errno));
+ goto done;
+ }
Hi Rich,
Sorry it took so long.
This all looks fine.
The only suggestion I can make is to include the "path"
in the three diagnostics above. I imagine it might otherwise
be hard to understand what the diagnostic is talking about otherwise.