qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
* [PATCH 0/6] tcg: Always implement neg and movcond
@ 2023-10-26  4:13 Richard Henderson
  2023-10-26  4:13 ` [PATCH 1/6] tcg/mips: Split out tcg_out_setcond_int Richard Henderson
                   ` (6 more replies)
  0 siblings, 7 replies; 8+ messages in thread
From: Richard Henderson @ 2023-10-26  4:13 UTC (permalink / raw)
  To: qemu-devel

Having opcodes always present means that we can remove some tests.
NOT and MOVCOND were *almost* always present anyway.

A close candidate is NOT.  The hiccup is s390x: except for the most
recent hardware revision, there is no single insn which can implement
the operation.

I experimented with replacements:

    i32: xilf r,-1             (6 bytes, requires R=X)
         lcr r,x; ahi r,-1     (6 bytes)
    i64: lcgr r,x; aghi r,-1   (8 bytes)

But both don't compare well with the current

         lghi tmp,-1           (4 bytes, shared)
         xgrk r,x,tmp          (4 bytes)

With the constant -1 managed by register allocation, it gets loaded
once and then reused between operations.

An alternative might be to assume the presence of all opcodes during
initial expansion and lower them later, after optimization.  I'm close
to doing just that for the more complex opcodes like deposit anyway,
because the expansion is too complex for the optimizer to do anything
sensible after constant propagation.


r~


Richard Henderson (6):
  tcg/mips: Split out tcg_out_setcond_int
  tcg/mips: Always implement movcond
  tcg: Remove TCG_TARGET_HAS_movcond_{i32,i64}
  tcg/mips: Implement neg opcodes
  tcg/loongarch64: Implement neg opcodes
  tcg: Remove TCG_TARGET_HAS_neg_{i32,i64}

 include/tcg/tcg-op-common.h      |  12 +-
 include/tcg/tcg-opc.h            |   8 +-
 include/tcg/tcg.h                |   2 -
 tcg/aarch64/tcg-target.h         |   4 -
 tcg/arm/tcg-target.h             |   2 -
 tcg/i386/tcg-target.h            |   4 -
 tcg/loongarch64/tcg-target.h     |   4 -
 tcg/mips/tcg-target.h            |   4 -
 tcg/ppc/tcg-target.h             |   4 -
 tcg/riscv/tcg-target.h           |   4 -
 tcg/s390x/tcg-target.h           |   4 -
 tcg/sparc64/tcg-target.h         |   4 -
 tcg/tci/tcg-target.h             |   4 -
 tcg/optimize.c                   |  15 +-
 tcg/tcg-op.c                     |  60 ++---
 tcg/tcg.c                        |  12 +-
 tcg/tci.c                        |   2 -
 tcg/loongarch64/tcg-target.c.inc |   9 +
 tcg/mips/tcg-target.c.inc        | 383 +++++++++++++------------------
 19 files changed, 192 insertions(+), 349 deletions(-)

-- 
2.34.1



^ permalink raw reply	[flat|nested] 8+ messages in thread

end of thread, other threads:[~2023-10-26  4:22 UTC | newest]

Thread overview: 8+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2023-10-26  4:13 [PATCH 0/6] tcg: Always implement neg and movcond Richard Henderson
2023-10-26  4:13 ` [PATCH 1/6] tcg/mips: Split out tcg_out_setcond_int Richard Henderson
2023-10-26  4:14 ` [PATCH 2/6] tcg/mips: Always implement movcond Richard Henderson
2023-10-26  4:14 ` [PATCH 3/6] tcg: Remove TCG_TARGET_HAS_movcond_{i32,i64} Richard Henderson
2023-10-26  4:14 ` [PATCH 4/6] tcg/mips: Implement neg opcodes Richard Henderson
2023-10-26  4:14 ` [PATCH 5/6] tcg/loongarch64: " Richard Henderson
2023-10-26  4:14 ` [PATCH 6/6] tcg: Remove TCG_TARGET_HAS_neg_{i32,i64} Richard Henderson
2023-10-26  4:20 ` [PATCH 0/6] tcg: Always implement neg and movcond Richard Henderson

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).