Bug 5579 – Segfault on first call to GC after starting new thread
Status
RESOLVED
Resolution
FIXED
Severity
regression
Priority
P2
Component
dmd
Product
D
Version
D2
Platform
x86
OS
Windows
Creation time
2011-02-14T17:46:00Z
Last change time
2011-02-14T22:24:29Z
Assigned to
nobody
Creator
dsimcha
Comments
Comment #0 by dsimcha — 2011-02-14T17:46:30Z
The following code segfaults on about 50% of runs. This happens in both 32-bit and 64-bit Linux and on Windows, regardless of how the -O, -inline, and -release flags are set.
import core.memory, core.thread;
void doNothing() {}
void main() {
auto t = new Thread(&doNothing);
t.start;
GC.collect();
}
Comment #1 by bugzilla — 2011-02-14T21:31:23Z
It faults on this code in druntime/src/rt/lifetime.d:
==================================
// we expect this to be called with the lock in place
extern(C) void rt_processGCMarks(void[] tls)
{
// called after the mark routine to eliminate block cache data when it
// might be ready to sweep
debug(PRINTF) printf("processing GC Marks, %x\n", tls.ptr);
auto cache = *cast(BlkInfo **)(tls.ptr + __blkcache_offset);
^^^^^^^^^^^^^^^^^^^^^^^^^^^
=================================
Both tls.ptr and __blkcache_offset are NULL.