building native JDK 1.5.0 and linprocfs witness reports

[Divacky Roman]

On Sat, Jul 08, 2006 at 08:27:30PM +0400, Yuri Pankov wrote:
> Hi.
> 
> I've problems compiling native JDK 1.5.0. Posting to this list as it
> seems related to linprocfs problems (linux-sun-jdk is used for
> bootstraping). Build fails early in the beginning with "javac"
> segmentation fault, and here's what I get in dmesg (lots of such
> messages):
> malloc(M_WAITOK) of "1024", forcing M_NOWAIT with the
> following non-sleepable locks held: exclusive sleep mutex vm object
> (standard object) r = 0 (0xffffff0063434578) locked
> @ /usr/src/sys/compat/linprocfs/linprocfs.c:835

this happens because the code locks the vm object which non-sleepable
mutex to protect it and then subsequent code calls malloc with WAITOK
which is forbidden in such cases.

pls, can you try attached patch and tell me whats printed if you reproduce
the problem? its not meant to fix anything just a info print

thnx, roman

Index: linprocfs.c
===================================================================
RCS file: /home/ncvs/src/sys/compat/linprocfs/linprocfs.c,v
retrieving revision 1.96
diff -u -r1.96 linprocfs.c
--- linprocfs.c	27 Jun 2006 20:11:58 -0000	1.96
+++ linprocfs.c	9 Jul 2006 09:14:55 -0000
@@ -832,6 +832,7 @@
 		ino = 0;
 		if (lobj) {
 			vp = lobj->handle;
+			printf("lobj type: %i\n", lobj->type);
 			VM_OBJECT_LOCK(lobj);
 			off = IDX_TO_OFF(lobj->size);
 			if (lobj->type == OBJT_VNODE && lobj->handle) {