Thread (27 messages) flat view 27 messages, 4 authors, 2007-09-13
STALE6930d

[PATCH 07/15] Don't return -ENOSYS as extra notes size if spufs is not loaded

From: Michael Ellerman <hidden>
Date: 2007-09-12 07:46:09
Also in: lkml
Subsystem: cell broadband engine architecture, linux for powerpc (32-bit and 64-bit), the rest · Maintainers: Madhavan Srinivasan, Linus Torvalds

Because the SPU coredump code might be built as part of a module (spufs),
we have a stub which is called by the coredump code, this routine then calls
into spufs if it's loaded.

Unfortunately the stub returns -ENOSYS if spufs is not loaded, which is
interpreted by the coredump code as an extra note size of -38 bytes. This
leads to a corrupt core dump.

If spufs is not loaded there will be no SPU ELF notes to write, and so the
extra notes size will be == 0.

Signed-off-by: Michael Ellerman <redacted>
---
 arch/powerpc/platforms/cell/spu_coredump.c |    8 ++++++--
 1 files changed, 6 insertions(+), 2 deletions(-)
diff --git a/arch/powerpc/platforms/cell/spu_coredump.c b/arch/powerpc/platforms/cell/spu_coredump.c
index 4fd37ff..656a8c5 100644
--- a/arch/powerpc/platforms/cell/spu_coredump.c
+++ b/arch/powerpc/platforms/cell/spu_coredump.c
@@ -31,15 +31,19 @@ static DEFINE_MUTEX(spu_coredump_mutex);
 
 int arch_notes_size(void)
 {
-	long ret;
+	int ret;
 
-	ret = -ENOSYS;
 	mutex_lock(&spu_coredump_mutex);
+
 	if (spu_coredump_calls && try_module_get(spu_coredump_calls->owner)) {
 		ret = spu_coredump_calls->arch_notes_size();
 		module_put(spu_coredump_calls->owner);
+	} else {
+		ret = 0;
 	}
+
 	mutex_unlock(&spu_coredump_mutex);
+
 	return ret;
 }
 
-- 
1.5.1.3.g7a33b
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help