On Thu, Apr 21, 2022 at 02:11:03PM +0200, Peter Krempa wrote:
On Wed, Apr 20, 2022 at 21:07:50 +0200, Victor Toso wrote:
> It doesn't need to fail the test if a reentrant function is found in
> a comment.
That is true, but your patch doesn't do exactly that.
If I add the following code:
@@ -83,6 +83,12 @@ virshCatchDisconnect(virConnectPtr conn,
const char *str = "unknown reason";
virErrorPtr error;
g_autofree char *uri = NULL;
+ long r;
+
+ r = random(); /* let's play a game */
+
+ if (r == 4)
+ abort();
It fails prior to this patch, but succeeds after it.
Yeah.
> Fixes:
> | ...
> | prohibit_nonreentrant
> | libvirt/ci/helper:33: default=os.getlogin(), # exempt from
syntax-check
> | libvirt/include/libvirt/libvirt-storage.h:223:
VIR_STORAGE_VOL_WIPE_ALG_PFITZNER7 = 6, /* 7-pass random (Since: v1.0.0) */
> | libvirt/include/libvirt/libvirt-storage.h:225:
VIR_STORAGE_VOL_WIPE_ALG_PFITZNER33 = 7, /* 33-pass random (Since: v1.0.0) */
> | libvirt/include/libvirt/libvirt-storage.h:227: VIR_STORAGE_VOL_WIPE_ALG_RANDOM
= 8, /* 1-pass random (Since: v1.0.0) */
Since it's very hard to parse C with regexes I suggest to
change
'pass random' to 'pass random data'
in the comments above to sidestep the issue.
Sure.
Thanks,
Victor