On 06/22/2011 06:20 PM, Andrew Haley wrote:
> I've found and fixed the bugs that were stopping gcj from working.
This is really impressive. Thanks for describing the details!
Cheers,
Niels
> Firstly, there was the "libgcc is a symlink instead of a linker
> script" bug that I've explained before.
>
>
> Then there was the libgcj/sysdep/arm/locks.h problem that broke
> locking. I already described that.
>
>
> Then I found a locking bug in the boehm garbage collector.
> GC_clear(), which releases a lock, was simply
>
> inline static void GC_clear(volatile unsigned int *addr) {
> /* Try to discourage gcc from moving anything past this. */
> __asm__ __volatile__(" " : : : "memory");
> *(addr) = 0;
> }
>
> but it needs a memory barrier instruction. The easiest way to fix
> this is to use a gcc builtin that generates the correct barrier:
>
> inline static void GC_clear(volatile unsigned int *addr) {
> __sync_lock_release(addr);
> }
>
> I also replaced GC_test_and_set() with a gcc builtin, although this
> isn't strictly necessary:
>
> inline static int GC_test_and_set(volatile unsigned int *addr) {
> return __sync_lock_test_and_set(addr, 1);
> }
>
>
> There was also a problem in libffi when generating a closure. The
> code that generates a trampoline looks like this:
>
> #define FFI_INIT_TRAMPOLINE(TRAMP,FUN,CTX) \
> ({ unsigned char *__tramp = (unsigned char*)(TRAMP); \
> unsigned int __fun = (unsigned int)(FUN); \
> unsigned int __ctx = (unsigned int)(CTX); \
> *(unsigned int*) &__tramp[0] = 0xe92d000f; /* stmfd sp!, {r0-r3} */
> \
> *(unsigned int*) &__tramp[4] = 0xe59f0000; /* ldr r0, [pc] */ \
> *(unsigned int*) &__tramp[8] = 0xe59ff000; /* ldr pc, [pc] */ \
> *(unsigned int*) &__tramp[12] = __ctx; \
> *(unsigned int*) &__tramp[16] = __fun; \
> __clear_cache((&__tramp[0]), (&__tramp[19])); \
> })
>
> Here, three instructions are laid down in memory, followed by some
> data fields. Once that has been done, __clear_cache() is called on
> the block of memory. Unfortunately, due to the requirement to placate
> SELinux, the same block of memory is mapped twice, once with RW
> mapping and once with RX mapping. Therefore, there must be two calls
> to __clear_cache(), one to flush the dcache, and one to flush the
> icache. Like this:
>
> #define FFI_INIT_TRAMPOLINE(TRAMP,FUN,CTX) \
> ({ unsigned char *__tramp = (unsigned char*)(TRAMP); \
> unsigned int __fun = (unsigned int)(FUN); \
> unsigned int __ctx = (unsigned int)(CTX); \
> unsigned char *insns = (unsigned char *)(CTX); \
> *(unsigned int*) &__tramp[0] = 0xe92d000f; /* stmfd sp!, {r0-r3} */
> \
> *(unsigned int*) &__tramp[4] = 0xe59f0000; /* ldr r0, [pc] */ \
> *(unsigned int*) &__tramp[8] = 0xe59ff000; /* ldr pc, [pc] */ \
> *(unsigned int*) &__tramp[12] = __ctx; \
> *(unsigned int*) &__tramp[16] = __fun; \
> __clear_cache((&__tramp[0]), (&__tramp[19])); /* Clear data mapping. */ \
> __clear_cache(insns, insns + 3 * sizeof (unsigned int)); \
> /* Clear instruction \
> mapping. */ \
> })
>
>
> Finally, libjava was segfaulting when an array class was finalized.
> This has apparently always happened, but on other platforms the
> segfault is caught and turned into a NullPointerException. All
> finalizers are called from an exception region that traps all
> exceptions and continues, so we never noticed.
>
> Fixed thusly:
>
> --- libjava/java/lang/natClass.cc 2010-06-11 05:09:01.000000000 -0400
> +++ prev/libjava/java/lang/natClass.cc 2011-06-21 06:09:22.000000000
> -0400
> @@ -668,6 +668,8 @@
> void
> java::lang::Class::finalize (void)
> {
> + // Array classes don't have an engine, and don't need to be finalized.
> + if (engine)
> engine->unregister(this);
> }
>
>
> After all this, gcj now seems to work. I'm now building OpenJDK.
>
> Andrew.
> _______________________________________________
> arm mailing list
> [email protected]
> https://admin.fedoraproject.org/mailman/listinfo/arm
_______________________________________________
arm mailing list
[email protected]
https://admin.fedoraproject.org/mailman/listinfo/arm
