- Purposefully add 6 debug prints to RtlpGetRegistryHandle, which, while ndis.sys is initializing, should BSOD the system. I have seen this "RtlpGetRegistryHandle" BSOD randomly during boot (But always with the same stack trace), and a local HAL patch managed to cause it 100% of the time. I know that the problem had nothing to do with the HAL or registry paths, because it came up before too when I added debug prints to fstub. These 6 prints 100% reproduce the problem on my machine with GCC 3.4.5. It is my hope it will do the same for everyone. This is a nasty bug, either stack related or some critical structure is being damaged. I am breaking trunk because it's been here for a long time, but now I found a way to repro it so that we can work together to try finding it. In a way, trunk was always broken, but this patch should force a freeze. Please help.

svn path=/trunk/; revision=24767
This commit is contained in:
Alex Ionescu 2006-11-16 07:09:37 +00:00
parent 8c3485a307
commit 0cb9e0fd78

View file

@ -94,6 +94,11 @@ RtlpGetRegistryHandle(IN ULONG RelativeTo,
}
/* And now append the path */
DPRINT1("I'm about to crash due to an overwrite problem, Alex thinks\n");
DPRINT1("I'm about to crash due to a overwrite problem, Alex thinks\n");
DPRINT1("I'm about to crash due to a overwrite problem, Alex thinks\n");
DPRINT1("I'm about to crash due to a overwrite problem, Alex thinks\n");
DPRINT1("I'm about to crash due to a overwrite problem, Alex thinks\n");
if (Path[0] == L'\\' && RelativeTo != RTL_REGISTRY_ABSOLUTE) Path++; // HACK!
Status = RtlAppendUnicodeToString(&KeyName, Path);
if (!NT_SUCCESS(Status)) return Status;