> Unfortunately, there's another problem. Both with and without your fix,
> seamonkey crashes when loading its default homepage on a PKU machine.
The faulting address is
$1 = (void *) 0x3b86ff63dfc <_sk_srcover_bgra_8888_sse2_lowp+688>
which is in a giant assembly file:
$ wc gfx/skia/skia/src/jumper/SkJumper_generated.S
83546 399499 6179831 gfx/skia/skia/src/jumper/SkJumper_generated.S
#0 thrkill () at /tmp/-:3
#1 0xf0c92e400598c794 in ?? ()
#2 0x000003b87086885b in nsProfileLock::FatalSignalHandler (signo=11,
info=0x7f7ffffb4810,
context=0x7f7ffffb4720)
at
/usr/ports/pobj/seamonkey-2.53.15/seamonkey-2.53.15/toolkit/profile/nsProfileLock.cpp:177
#3 <signal handler called>
#4 0x000003b86ff2fd93 in _sk_black_color_hsw () from
/usr/local/lib/seamonkey/libxul.so.47.0
#5 0x000003b86ff2fc1d in _sk_start_pipeline_hsw () from
/usr/local/lib/seamonkey/libxul.so.47.0
#6 0x000003b870037870 in SkRasterPipeline::run (this=0x0, x=4090687061393, y=0,
w=<optimized out>, h=<optimized out>)
It appears that the code in this .S file is both read and executed by
seamonkey, so switching to .rodata doesn't work. Given the experience
in node and other browsers with similar embedded files, I think it is
more prudent to apply the big hammer for now.
Index: Makefile
===================================================================
RCS file: /cvs/ports/www/seamonkey/Makefile,v
retrieving revision 1.284
diff -u -p -r1.284 Makefile
--- Makefile 26 Jan 2023 20:46:34 -0000 1.284
+++ Makefile 26 Jan 2023 21:47:15 -0000
@@ -1,5 +1,7 @@
BROKEN-aarch64 = third_party/rust/libc/src/lib.rs: error[E0412]: cannot
find type 'c_long' in this scope
+USE_NOEXECONLY = Yes
+
COMMENT-main = Mozilla application suite
COMMENT-lightning = Mozilla Seamonkey calendar extension
ONLY_FOR_ARCHS = amd64 i386 aarch64
@@ -13,7 +15,7 @@ MULTI_PACKAGES = -main -lightning
PKGNAME-main = ${PKGNAME}
PKGNAME-lightning = lightning-seamonkey-5.8.15
EPOCH-lightning = 0
-REVISION-main = 0
+REVISION-main = 1
HOMEPAGE = https://www.seamonkey-project.org/
