diff options
author | Björn Gustavsson <[email protected]> | 2018-12-03 10:23:15 +0100 |
---|---|---|
committer | Björn Gustavsson <[email protected]> | 2018-12-13 10:31:22 +0100 |
commit | 3691c6ab52ace5a3f0c1ddf51b7ab6e0b9abb385 (patch) | |
tree | 52a10fc4ec361d4dca69186d9bfc1d30d5d52b47 /erts/emulator/beam/ops.tab | |
parent | f19a52e92ae56b8e5248064db56dc38c0d7dff69 (diff) | |
download | otp-3691c6ab52ace5a3f0c1ddf51b7ab6e0b9abb385.tar.gz otp-3691c6ab52ace5a3f0c1ddf51b7ab6e0b9abb385.tar.bz2 otp-3691c6ab52ace5a3f0c1ddf51b7ab6e0b9abb385.zip |
Simplify GC BIFs
Summary: This commit simplifies the implementation of the "GC BIFs" so
that they no longer need to do a garbage collection, removing duplicate
code for all GC BIFs in the runtime system, as well as potentially
reducing the size of the loaded BEAM code by using shorter
instructions calling those BIFs.
A GC BIF is a guard BIF that will do a garbage
collection if it needs to build anything on the heap.
For example, `abs/1` is a GC BIF because it might need to
allocate space on the heap (if the result is a floating point
number or the resulting integer is a bignum).
Before R12, a guard BIF (such as `abs/1`) that need to allocate
heap space would allocate outside of process's main heap, in
a heap fragment.
GC BIFs were introduced in R12B to support literals. During garbage
collection it become necessary to quickly test whether a term was
a literal. To make the check simple, guards BIFs were no longer
allowed to create heap fragments. Instead GC BIFs were introduced.
In OTP 19, the implementation of literals was changed to support
storing messages in heap fragments outside of the main heap for a
process. That change again made it allowed for guard BIFs to create
heap fragments when they need to build terms on the heap.
It would even be possible for the guard BIFs to build directly
on the main heap if there is room there, because the compiler
assumes that a new `test_heap/2` instruction must be emitted
when building anything after calling a GC BIF. (We don't do that
in this commit; see below.)
This commit simplifies the implementation of the GC BIFs in
the runtime system.
Each GC BIF had a dual implementation: one that was used when the GC
BIF was called directly and one used when it was called via
`apply/3`. For example, `abs/1` was implemented in `abs_1()` and
`erts_gc_abs_1()`. This commit removes the GC version of each BIF. The
other version that allocates heap space using `HAlloc()` is updated to
use the new `HeapFragOnlyAlloc()` macro that will allocate heap
space in a heap fragment outside of the main heap.
Because the BIFs will allocate outside of the main heap, the same
`bif` instructions used by nonbuilding BIFs can be used for the
(former) GC BIFs. Those instructions don't use the macros that save
and restore the heap and stack pointers (SWAPOUT/SWAPIN). If the
former GC BIFs would build on the main heap, either new instructions
would be needed, or SWAPOUT/SWAPIN instructions would need to be added
to the `bif` instructions.
Instructions that call the former GC BIFs don't need the operand
that specifies the number of live X registers. Therefore, the
instructions that call the BIFs are usually one word shorter.
Diffstat (limited to 'erts/emulator/beam/ops.tab')
-rw-r--r-- | erts/emulator/beam/ops.tab | 108 |
1 files changed, 49 insertions, 59 deletions
diff --git a/erts/emulator/beam/ops.tab b/erts/emulator/beam/ops.tab index 349034e8ac..fbed2e56e1 100644 --- a/erts/emulator/beam/ops.tab +++ b/erts/emulator/beam/ops.tab @@ -1002,10 +1002,11 @@ bif1 Fail Bif=u$bif:erlang:get/1 Src=s Dst=d => gen_get(Src, Dst) bif2 Jump=j u$bif:erlang:element/2 S1=s S2=xy Dst=d => gen_element(Jump, S1, S2, Dst) -bif1 p Bif S1 Dst => bif1_body Bif S1 Dst +bif1 p Bif S1 Dst => i_bif1_body Bif S1 Dst +bif1 Fail=f Bif S1 Dst => i_bif1 Fail Bif S1 Dst -bif2 p Bif S1 S2 Dst => i_bif2_body Bif S1 S2 Dst -bif2 Fail Bif S1 S2 Dst => i_bif2 Fail Bif S1 S2 Dst +bif2 p Bif S1 S2 Dst => i_bif2_body Bif S1 S2 Dst +bif2 Fail=f Bif S1 S2 Dst => i_bif2 Fail Bif S1 S2 Dst i_get_hash c I d i_get s d @@ -1023,10 +1024,12 @@ i_fast_element xy j? I d i_element xy j? s d -bif1 f? b s d -bif1_body b s d +i_bif1 f? b s d +i_bif1_body b s d i_bif2 f? b s s d i_bif2_body b s s d +i_bif3 f? b s s s d +i_bif3_body b s s s d # # Internal calls. @@ -1499,80 +1502,80 @@ gc_bif2 Fail Live u$bif:erlang:sminus/2 S1 S2 Dst => \ # # Optimize addition and subtraction of small literals using -# the i_increment/4 instruction (in bodies, not in guards). +# the i_increment/3 instruction (in bodies, not in guards). # gen_plus p Live Int=i Reg=d Dst => \ - gen_increment(Reg, Int, Live, Dst) + gen_increment(Reg, Int, Dst) gen_plus p Live Reg=d Int=i Dst => \ - gen_increment(Reg, Int, Live, Dst) + gen_increment(Reg, Int, Dst) gen_minus p Live Reg=d Int=i Dst | negation_is_small(Int) => \ - gen_increment_from_minus(Reg, Int, Live, Dst) + gen_increment_from_minus(Reg, Int, Dst) # -# GCing arithmetic instructions. +# Arithmetic instructions. # -gen_plus Fail Live S1 S2 Dst => i_plus S1 S2 Fail Live Dst +gen_plus Fail Live S1 S2 Dst => i_plus S1 S2 Fail Dst -gen_minus Fail Live S1 S2 Dst => i_minus S1 S2 Fail Live Dst +gen_minus Fail Live S1 S2 Dst => i_minus S1 S2 Fail Dst gc_bif2 Fail Live u$bif:erlang:stimes/2 S1 S2 Dst => \ - i_times Fail Live S1 S2 Dst + i_times Fail S1 S2 Dst gc_bif2 Fail Live u$bif:erlang:div/2 S1 S2 Dst => \ - i_m_div Fail Live S1 S2 Dst + i_m_div Fail S1 S2 Dst gc_bif2 Fail Live u$bif:erlang:intdiv/2 S1 S2 Dst => \ - i_int_div Fail Live S1 S2 Dst + i_int_div Fail S1 S2 Dst gc_bif2 Fail Live u$bif:erlang:rem/2 S1 S2 Dst => \ - i_rem S1 S2 Fail Live Dst + i_rem S1 S2 Fail Dst gc_bif2 Fail Live u$bif:erlang:bsl/2 S1 S2 Dst => \ - i_bsl S1 S2 Fail Live Dst + i_bsl S1 S2 Fail Dst gc_bif2 Fail Live u$bif:erlang:bsr/2 S1 S2 Dst => \ - i_bsr S1 S2 Fail Live Dst + i_bsr S1 S2 Fail Dst gc_bif2 Fail Live u$bif:erlang:band/2 S1 S2 Dst => \ - i_band S1 S2 Fail Live Dst + i_band S1 S2 Fail Dst gc_bif2 Fail Live u$bif:erlang:bor/2 S1 S2 Dst => \ - i_bor Fail Live S1 S2 Dst + i_bor Fail S1 S2 Dst gc_bif2 Fail Live u$bif:erlang:bxor/2 S1 S2 Dst => \ - i_bxor Fail Live S1 S2 Dst + i_bxor Fail S1 S2 Dst -gc_bif1 Fail I u$bif:erlang:bnot/1 Src Dst=d => i_int_bnot Fail Src I Dst +gc_bif1 Fail Live u$bif:erlang:bnot/1 Src Dst=d => i_int_bnot Fail Src Dst -i_increment rxy W t d +i_increment rxy W d -i_plus x xy j? t d -i_plus s s j? t d +i_plus x xy j? d +i_plus s s j? d -i_minus x x j? t d -i_minus s s j? t d +i_minus x x j? d +i_minus s s j? d -i_times j? t s s d +i_times j? s s d -i_m_div j? t s s d -i_int_div j? t s s d +i_m_div j? s s d +i_int_div j? s s d -i_rem x x j? t d -i_rem s s j? t d +i_rem x x j? d +i_rem s s j? d -i_bsl s s j? t d -i_bsr s s j? t d +i_bsl s s j? d +i_bsr s s j? d -i_band x c j? t d -i_band s s j? t d +i_band x c j? d +i_band s s j? d -i_bor j? I s s d -i_bxor j? I s s d +i_bor j? s s d +i_bxor j? s s d -i_int_bnot Fail Src=c Live Dst => move Src x | i_int_bnot Fail x Live Dst +i_int_bnot Fail Src=c Dst => move Src x | i_int_bnot Fail x Dst -i_int_bnot j? S t d +i_int_bnot j? S d # # Old guard BIFs that creates heap fragments are no longer allowed. @@ -1587,27 +1590,14 @@ bif1 Fail u$bif:erlang:trunc/1 s d => too_old_compiler # # Guard BIFs. # -gc_bif1 Fail I Bif Src Dst => \ - gen_guard_bif1(Fail, I, Bif, Src, Dst) +gc_bif1 p Live Bif Src Dst => i_bif1_body Bif Src Dst +gc_bif1 Fail=f Live Bif Src Dst => i_bif1 Fail Bif Src Dst -gc_bif2 Fail I Bif S1 S2 Dst => \ - gen_guard_bif2(Fail, I, Bif, S1, S2, Dst) +gc_bif2 p Live Bif S1 S2 Dst => i_bif2_body Bif S1 S2 Dst +gc_bif2 Fail=f Live Bif S1 S2 Dst => i_bif2 Fail Bif S1 S2 Dst -gc_bif3 Fail I Bif S1 S2 S3 Dst => \ - gen_guard_bif3(Fail, I, Bif, S1, S2, S3, Dst) - -i_gc_bif1 j? W s t? d - -i_gc_bif2 j? W t? s s d - -ii_gc_bif3/7 - -# A specific instruction can only have 6 operands, so we must -# pass one of the arguments in an x register. -ii_gc_bif3 Fail Bif Live S1 S2 S3 Dst => \ - move S1 x | i_gc_bif3 Fail Bif Live S2 S3 Dst - -i_gc_bif3 j? W t? s s d +gc_bif3 p Live Bif S1 S2 S3 Dst => i_bif3_body Bif S1 S2 S3 Dst +gc_bif3 Fail=f Live Bif S1 S2 S3 Dst => i_bif3 Fail Bif S1 S2 S3 Dst # # The following instruction is specially handled in beam_load.c |