From 8ef6873e0e14309a1715a252a650bab0ae1a33ef Mon Sep 17 00:00:00 2001 From: Josh Haberman Date: Sun, 20 Mar 2011 13:13:51 -0700 Subject: upb_stream: all callbacks registered ahead-of-time. This is a significant change to the upb_stream protocol, and should hopefully be the last significant change. All callbacks are now registered ahead-of-time instead of having delegated callbacks registered at runtime, which makes it much easier to aggressively optimize ahead-of-time (like with a JIT). Other impacts of this change: - You no longer need to have loaded descriptor.proto as a upb_def to load other descriptors! This means the special-case code we used for bootstrapping is no longer necessary, and we no longer need to link the descriptor for descriptor.proto into upb. - A client can now register any upb_value as what will be delivered to their value callback, not just a upb_fielddef*. This should allow for other clients to get more bang out of the streaming decoder. This change unfortunately causes a bit of a performance regression -- I think largely due to highly suboptimal code that GCC generates when structs are returned by value. See: http://blog.reverberate.org/2011/03/19/when-a-compilers-slow-code-actually-bites-you/ On the other hand, once we have a JIT this should no longer matter. Performance numbers: plain.parsestream_googlemessage1.upb_table: 374 -> 396 (5.88) plain.parsestream_googlemessage2.upb_table: 616 -> 449 (-27.11) plain.parsetostruct_googlemessage1.upb_table_byref: 268 -> 269 (0.37) plain.parsetostruct_googlemessage1.upb_table_byval: 215 -> 204 (-5.12) plain.parsetostruct_googlemessage2.upb_table_byref: 307 -> 281 (-8.47) plain.parsetostruct_googlemessage2.upb_table_byval: 297 -> 272 (-8.42) omitfp.parsestream_googlemessage1.upb_table: 423 -> 410 (-3.07) omitfp.parsestream_googlemessage2.upb_table: 679 -> 483 (-28.87) omitfp.parsetostruct_googlemessage1.upb_table_byref: 287 -> 282 (-1.74) omitfp.parsetostruct_googlemessage1.upb_table_byval: 226 -> 219 (-3.10) omitfp.parsetostruct_googlemessage2.upb_table_byref: 315 -> 298 (-5.40) omitfp.parsetostruct_googlemessage2.upb_table_byval: 297 -> 287 (-3.37) --- src/upb_decoder_x64.asm | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'src/upb_decoder_x64.asm') diff --git a/src/upb_decoder_x64.asm b/src/upb_decoder_x64.asm index 032ea86..c417644 100644 --- a/src/upb_decoder_x64.asm +++ b/src/upb_decoder_x64.asm @@ -34,8 +34,8 @@ SECTION .text %define BUF rbx ; const char *p, current buf position. %define END rbp ; const char *end, where the buf ends (either submsg end or buf end) %define STRING r12 ; unused -%define FIELDDEF r13 ; upb_fielddef *f, needs to be preserved across varint decoding call. -%define CALLBACK r14 +%define FVAL r13 ; upb_value fval, needs to be preserved across varint decoding call. +%define UNUSED r14 %define CLOSURE r15 ; Stack layout: *tableptr, uint32_t maxfield_times_8 @@ -57,10 +57,10 @@ SECTION .text ; path that goes into a tight loop if the encoding was packed). ; - check_6: the field is not a group or a message (or string, TODO) ; (this could be relaxed, but due to delegation it's a bit tricky). -; - if the value is a string, the entire string is available in +; - check_7: if the value is a string, the entire string is available in ; the buffer, and our cached string object can be recycled, and ; our string object already references the source buffer, so -; absolutely no refcount twiddling is required. (check_7) +; absolutely no refcount twiddling is required. %macro decode_and_dispatch_ 0 @@ -78,7 +78,7 @@ align 16 ; Decode a 1 or 2-byte varint -> eax. mov cl, byte [BUF] lea rdi, [BUF+1] - movzx rax, cl ; Need all of rax since we're doing a 64-bit lea later. + movzx eax, cl and eax, 0x7f test cl, cl jns .one_byte_tag ; Should be predictable if fields are in order. -- cgit v1.2.3