From 857ac1e817808f4b6bf985679162d0e3d709e5b5 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Thu, 26 Feb 2009 05:35:15 -0800 Subject: mesa: Resurrect SPARC asm code. This rewrites the sparc GLAPI code so that it's PIC friendly and works with all of the TLS/PTHREADS/64-bit/32-bit combinations properly. As a result we can turn SPARC asm back on. Currently it's only enabled on Linux, as that's the only place where I can test this stuff out. For the moment the cliptest SPARC asm routines are disabled as they are non-working. The problem is that they use register %g7 as a temporary which is where the threading libraries store the thread pointer on SPARC. I will fix that code up in a future change as it's a pretty important routine to optimize. Like x86 we do the runtime patch as a pthread once-invoked initializer in init_glapi_relocs(). Unlike x86, however, our GLAPI stubs on SPARC are just two instruction sequences that branch to a trampoline and put the GLAPI offset into a register. The trampoline is what we run-time patch. The stubs thus all look like: glFoo: ba __glapi_sparc_foo_stub sethi GLAPI_OFFSET(glFOO) * PTR_SIZE, %g3 This actually makes generate_entrypoint() a lot simpler on SPARC. For this case in generate_entrypoint() we generate stubs using a 'call' instead of the 'ba' above to make sure it can reach. In order to get a proper tail call going here, in the unpatched case, we do several tricks. To get the current PC, for example, we save the return address register into a temporary, do a call, save the return address register written by the call to another temporary, then restore the original return address register value. This is to avoid having to allocate a stack frame. This is necessary for PIC address formation. This new GLAPI scheme lets us get rid of the ugly SPARC GLAPI hacks in __glXInitialize() and one_time_init(). Signed-off-by: David S. Miller --- src/mesa/glapi/glapi_getproc.c | 78 ++++++++++++------------------------------ 1 file changed, 22 insertions(+), 56 deletions(-) (limited to 'src/mesa/glapi/glapi_getproc.c') diff --git a/src/mesa/glapi/glapi_getproc.c b/src/mesa/glapi/glapi_getproc.c index 4adc1aece0..764e2ef98e 100644 --- a/src/mesa/glapi/glapi_getproc.c +++ b/src/mesa/glapi/glapi_getproc.c @@ -259,53 +259,30 @@ generate_entrypoint(GLuint functionOffset) } return (_glapi_proc) code; -#elif defined(USE_SPARC_ASM) - -#ifdef __arch64__ - static const unsigned int insn_template[] = { - 0x05000000, /* sethi %uhi(_glapi_Dispatch), %g2 */ - 0x03000000, /* sethi %hi(_glapi_Dispatch), %g1 */ - 0x8410a000, /* or %g2, %ulo(_glapi_Dispatch), %g2 */ - 0x82106000, /* or %g1, %lo(_glapi_Dispatch), %g1 */ - 0x8528b020, /* sllx %g2, 32, %g2 */ - 0xc2584002, /* ldx [%g1 + %g2], %g1 */ - 0x05000000, /* sethi %hi(8 * glapioffset), %g2 */ - 0x8410a000, /* or %g2, %lo(8 * glapioffset), %g2 */ - 0xc6584002, /* ldx [%g1 + %g2], %g3 */ - 0x81c0c000, /* jmpl %g3, %g0 */ - 0x01000000 /* nop */ +#elif defined(USE_SPARC_ASM) && (defined(PTHREADS) || defined(GLX_USE_TLS)) + static const unsigned int template[] = { + 0x07000000, /* sethi %hi(0), %g3 */ + 0x8210000f, /* mov %o7, %g1 */ + 0x40000000, /* call */ + 0x9e100001, /* mov %g1, %o7 */ }; +#ifdef GLX_USE_TLS + extern unsigned int __glapi_sparc_tls_stub; + unsigned long call_dest = (unsigned long ) &__glapi_sparc_tls_stub; #else - static const unsigned int insn_template[] = { - 0x03000000, /* sethi %hi(_glapi_Dispatch), %g1 */ - 0xc2006000, /* ld [%g1 + %lo(_glapi_Dispatch)], %g1 */ - 0xc6006000, /* ld [%g1 + %lo(4*glapioffset)], %g3 */ - 0x81c0c000, /* jmpl %g3, %g0 */ - 0x01000000 /* nop */ - }; -#endif /* __arch64__ */ - unsigned int *code = (unsigned int *) malloc(sizeof(insn_template)); - unsigned long glapi_addr = (unsigned long) &_glapi_Dispatch; + extern unsigned int __glapi_sparc_pthread_stub; + unsigned long call_dest = (unsigned long ) &__glapi_sparc_pthread_stub; +#endif + unsigned int *code = (unsigned int *) malloc(sizeof(template)); if (code) { - memcpy(code, insn_template, sizeof(insn_template)); - -#ifdef __arch64__ - code[0] |= (glapi_addr >> (32 + 10)); - code[1] |= ((glapi_addr & 0xffffffff) >> 10); + code[0] = template[0] | (functionOffset & 0x3fffff); + code[1] = template[1]; __glapi_sparc_icache_flush(&code[0]); - code[2] |= ((glapi_addr >> 32) & ((1 << 10) - 1)); - code[3] |= (glapi_addr & ((1 << 10) - 1)); + code[2] = template[2] | + (((call_dest - ((unsigned long) &code[2])) + >> 2) & 0x3fffffff); + code[3] = template[3]; __glapi_sparc_icache_flush(&code[2]); - code[6] |= ((functionOffset * 8) >> 10); - code[7] |= ((functionOffset * 8) & ((1 << 10) - 1)); - __glapi_sparc_icache_flush(&code[6]); -#else - code[0] |= (glapi_addr >> 10); - code[1] |= (glapi_addr & ((1 << 10) - 1)); - __glapi_sparc_icache_flush(&code[0]); - code[2] |= (functionOffset * 4); - __glapi_sparc_icache_flush(&code[2]); -#endif /* __arch64__ */ } return (_glapi_proc) code; #else @@ -337,21 +314,10 @@ fill_in_entrypoint_offset(_glapi_proc entrypoint, GLuint offset) #endif #elif defined(USE_SPARC_ASM) - - /* XXX this hasn't been tested! */ unsigned int *code = (unsigned int *) entrypoint; -#ifdef __arch64__ - code[6] = 0x05000000; /* sethi %hi(8 * glapioffset), %g2 */ - code[7] = 0x8410a000; /* or %g2, %lo(8 * glapioffset), %g2 */ - code[6] |= ((offset * 8) >> 10); - code[7] |= ((offset * 8) & ((1 << 10) - 1)); - __glapi_sparc_icache_flush(&code[6]); -#else /* __arch64__ */ - code[2] = 0xc6006000; /* ld [%g1 + %lo(4*glapioffset)], %g3 */ - code[2] |= (offset * 4); - __glapi_sparc_icache_flush(&code[2]); -#endif /* __arch64__ */ - + code[0] &= ~0x3fffff; + code[0] |= (offset * sizeof(void *)) & 0x3fffff; + __glapi_sparc_icache_flush(&code[0]); #else /* an unimplemented architecture */ -- cgit v1.2.3