OpenJDK in core-updates

OpenJDK in core-updates

[Andreas Enge]

Hello,

something we have done recently in core-updates breaks the build of OpenJDK,
already at version 9. If someone could have a look, that would be great!

Andreas


/tmp/guix-build-openjdk-9.181.drv-0/source/hotspot/src/share/vm/opto/lcm.cpp: In function ‘bool accesses_heap_base_zone(Node*)’:
/tmp/guix-build-openjdk-9.181.drv-0/source/hotspot/src/share/vm/opto/lcm.cpp:42:35: error: ordered comparison of pointer with integer zero (‘address’ {aka ‘unsigned char*’} and ‘int’)
   42 |   if (Universe::narrow_oop_base() > 0) { // Implies UseCompressedOops.
      |       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~
make[3]: *** [lib/CompileJvm.gmk:209: /tmp/guix-build-openjdk-9.181.drv-0/source/build/linux-x86_64-normal-server-release/hotspot/variant-server/libjvm/objs/lcm.o] Error 1

Hm, it might be enough to replace "> 0" by "!= NULL" ?

Re: OpenJDK in core-updates

[Andreas Enge]

Am Wed, Apr 05, 2023 at 12:19:44PM +0200 schrieb Andreas Enge:
> Hm, it might be enough to replace "> 0" by "!= NULL" ?

Good guess, this is done in openjdk@10. I will give it a try.

Andreas

Re: OpenJDK in core-updates

[Andreas Enge]

Next problem, hopefully fixed by the following patch:

diff -u -r openjdk-9.alt/hotspot/test/runtime/StackGuardPages/exeinvoke.c openjdk-9/hotspot/test/runtime/StackGuardPages/exeinvoke.c
--- openjdk-9.alt/hotspot/test/runtime/StackGuardPages/exeinvoke.c	2023-04-05 13:46:58.689964892 +0200
+++ openjdk-9/hotspot/test/runtime/StackGuardPages/exeinvoke.c	2023-04-05 14:15:52.435613325 +0200
@@ -67,8 +67,17 @@
   longjmp(context, 1);
 }

+static char* altstack = NULL;
+
 void set_signal_handler() {
-  static char altstack[SIGSTKSZ];
+  if (altstack == NULL) {
+    // Dynamically allocated in case SIGSTKSZ is not constant
+    altstack = (char*)malloc(SIGSTKSZ);
+    if (altstack == NULL) {
+      fprintf(stderr, "Test ERROR. Unable to malloc altstack space\n");
+      exit(7);
+    }
+  }

   stack_t ss = {
     .ss_size = SIGSTKSZ,

This is related to a change in glibc, and we have seen it in other places.
It is not new, so I do not quite understand how openjdk@9 compiled a few
weeks ago on core-updates and suddenly stopped to compile. At least openjdk@10
is also affected, and maybe later versions.

Andreas

Re: OpenJDK in core-updates

[Andreas Enge]

Am Wed, Apr 05, 2023 at 02:26:07PM +0200 schrieb Andreas Enge:
> Next problem, hopefully fixed by the following patch:

Funny observation, the problems appear in even releases... So OpenJDK 12,
14 and 16 contain bugs that were fixed in 11, 13 and 15! As far as I can
tell, this is fixed now on core-updates.

Andreas