Summary: | Valgrind reports use of uninitialised variables | ||
---|---|---|---|
Product: | GhostPCL | Reporter: | Robin Watts <robin.watts> |
Component: | PCL interpreter | Assignee: | Henry Stiles <henry.stiles> |
Status: | RESOLVED FIXED | ||
Severity: | normal | ||
Priority: | P4 | ||
Version: | master | ||
Hardware: | PC | ||
OS: | Linux | ||
Customer: | Word Size: | --- |
Description
Robin Watts
2011-02-20 17:30:02 UTC
Technically this is a don't care UMR - the cap is set to 0 in pcursor.c:pcursort_reaset() which must be run before the cap is accessed in a useful manner. But it is now fixed in 12176. I'll leave the bug open, I don't know if valgrind error is the only issue or was suspected to be a symptom of something wrong. Revision 12176 does NOT stop the UMR, at least on my 32bit linux VMware image. The valgrind error given now, is actually different than before - control appears to arrive in pcl_set_cap_y with pcs->vmi_cp being undefined. Setting pcs->vmi_cp = 0; at the end of pcl_init_state appears to satisfy valgrind, but I have no idea if this is actually appropriate or not. (In reply to comment #3) > The valgrind error given now, is actually different than before - control > appears to arrive in pcl_set_cap_y with pcs->vmi_cp being undefined. > > Setting pcs->vmi_cp = 0; at the end of pcl_init_state appears to satisfy > valgrind, but I have no idea if this is actually appropriate or not. That I cannot reproduce, just to verify you mean: coord vmi_cp = pcs->vmi_cp; pcursor.c:258. Right? Fixed in revision 12181. Thanks. |