Check that the value matches opcode_wregs
Signed-off-by: Taylor Simpson <ltaylorsimpson@gmail.com>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20240307032327.4799-5-ltaylorsimpson@gmail.com>
Signed-off-by: Brian Cain <bcain@quicinc.com>
Check that the value matches opcode_reginfo/opcode_wregs
Signed-off-by: Taylor Simpson <ltaylorsimpson@gmail.com>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20240307032327.4799-4-ltaylorsimpson@gmail.com>
Signed-off-by: Brian Cain <bcain@quicinc.com>
Check that the value matches opcode_reginfo
Signed-off-by: Taylor Simpson <ltaylorsimpson@gmail.com>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20240307032327.4799-3-ltaylorsimpson@gmail.com>
Signed-off-by: Brian Cain <bcain@quicinc.com>
Signed-off-by: Taylor Simpson <ltaylorsimpson@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20240307032327.4799-2-ltaylorsimpson@gmail.com>
Signed-off-by: Brian Cain <bcain@quicinc.com>
Currently, we pass env to every generated helper. When the semantics of
the instruction only depend on the arguments, this is unnecessary and
adds extra overhead to the helper call.
We add the TCG_CALL_NO_RWG_SE flag to any non-HVX helpers that don't get
the ptr to env.
The A2_nop and SA1_setin1 instructions end up with no arguments. This
results in a "old-style function definition" error from the compiler, so
we write overrides for them.
With this change, the number of helpers with env argument is
idef-parser enabled: 329 total, 23 with env
idef-parser disabled: 1543 total, 550 with env
Signed-off-by: Taylor Simpson <ltaylorsimpson@gmail.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Tested-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20240214042726.19290-4-ltaylorsimpson@gmail.com>
Signed-off-by: Brian Cain <bcain@quicinc.com>
Rather than reading SP from the env, pass it explicitly
Signed-off-by: Taylor Simpson <ltaylorsimpson@gmail.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Tested-by: Anton Johansson <anjo@rev.ng>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20240214042726.19290-3-ltaylorsimpson@gmail.com>
Signed-off-by: Brian Cain <bcain@quicinc.com>
Rather than reading P0 from the env, pass it explicitly
Signed-off-by: Taylor Simpson <ltaylorsimpson@gmail.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Tested-by: Anton Johansson <anjo@rev.ng>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20240214042726.19290-2-ltaylorsimpson@gmail.com>
Signed-off-by: Brian Cain <bcain@quicinc.com>
Look for read-after-write instead of overlap of reads and writes
HVX instructions with helpers have pass-by-reference semantics, so
we check for overlaps of reads and writes within the same instruction.
Signed-off-by: Taylor Simpson <ltaylorsimpson@gmail.com>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20240201103340.119081-4-ltaylorsimpson@gmail.com>
Signed-off-by: Brian Cain <bcain@quicinc.com>
We divide gen_analyze_funcs.py into 3 phases
Declare the operands
Analyze the register reads
Analyze the register writes
We also create special versions of ctx_log_*_read for new operands
Check that the operand is written before the read
This is a precursor to improving the analysis for short-circuiting
the packet semantics in a subsequent commit
Signed-off-by: Taylor Simpson <ltaylorsimpson@gmail.com>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20240201103340.119081-2-ltaylorsimpson@gmail.com>
Signed-off-by: Brian Cain <bcain@quicinc.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20240503072014.24751-10-philmd@linaro.org>
Prepare for pcrel by not modifying cpu_pc before use,
in the case of JSR.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20240503072014.24751-9-philmd@linaro.org>
Richard Henderson explained on IRC:
bcond_internal() used to insist that both branch
destination and branch fallthrough are use_goto_tb;
if not, we'd use movcond to compute an indirect jump.
But it's perfectly fine for e.g. the branch fallthrough
to use_goto_tb, and the branch destination to use
an indirect branch.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240424234436.995410-4-richard.henderson@linaro.org>
[PMD: Split bigger patch, part 4/5]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20240503072014.24751-7-philmd@linaro.org>
Trivial change to make next commits easier to understand.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240424234436.995410-4-richard.henderson@linaro.org>
[PMD: Split bigger patch, part 3/5]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20240503072014.24751-6-philmd@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20240503072014.24751-3-philmd@linaro.org>
ALPHA_CPU has a dynamic object type assert, which is
unnecessary considering that these are all class hooks.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20240503072014.24751-2-philmd@linaro.org>
We only support the most recent two versions of macOS (currently
macOS 13 Ventura and macOS 14 Sonoma), and our ui/cocoa.m code
already assumes at least macOS 12 Monterey or better, because it uses
NSScreen safeAreaInsets, which is 12.0-or-newer.
Remove the ifdefs that were providing backwards compatibility for
building on 10.12 and earlier versions.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240502142904.62644-1-peter.maydell@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
To easily compare with the SH4 manual, rename:
REG(B11_8) -> Rn
REG(B7_4) -> Rm
t0 -> result
Mention how underflow is calculated.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20240430163125.77430-5-philmd@linaro.org>
To easily compare with the SH4 manual, rename:
REG(B11_8) -> Rn
REG(B7_4) -> Rm
t0 -> result
Mention how overflow is calculated.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Message-Id: <20240430163125.77430-4-philmd@linaro.org>
The documentation says:
SUBV Rm, Rn Rn - Rm -> Rn, underflow -> T
The overflow / underflow can be calculated as:
T = ((Rn ^ Rm) & (Result ^ Rn)) >> 31
However we were using the incorrect:
T = ((Rn ^ Rm) & (Result ^ Rm)) >> 31
Fix by using the Rn register instead of Rm.
Add tests provided by Paul Cercueil.
Cc: qemu-stable@nongnu.org
Fixes: ad8d25a11f ("target-sh4: implement addv and subv using TCG")
Reported-by: Paul Cercueil <paul@crapouillou.net>
Suggested-by: Paul Cercueil <paul@crapouillou.net>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2318
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Message-Id: <20240430163125.77430-3-philmd@linaro.org>
"plugin_mask" was renamed as "event_mask" in commit c006147122
("plugins: create CPUPluginState and migrate plugin_mask").
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240427155714.53669-3-philmd@linaro.org>
"exec/ram_addr.h" shouldn't be used with user emulation.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240427155714.53669-4-philmd@linaro.org>
All user emulation headers are now under include/user/.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240428221450.26460-3-philmd@linaro.org>
Keep all user emulation headers under the same user/ directory.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240428221450.26460-2-philmd@linaro.org>
Keep all user emulation headers under the same user/ directory.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240503125202.35667-1-philmd@linaro.org>
Commit 1ad2134f91 ("Hardware convenience library") extracted
"cpu-common.h" from "cpu-all.h", which uses the LGPL-2.1+ license.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240427155714.53669-5-philmd@linaro.org>
Benchmark each acceleration function vs an aligned buffer of zeros.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Because non-embedded aarch64 is expected to have AdvSIMD enabled, merely
double-check with the compiler flags for __ARM_NEON and don't bother with
a runtime check. Otherwise, model the loop after the x86 SSE2 function.
Use UMAXV for the vector reduction. This is 3 cycles on cortex-a76 and
2 cycles on neoverse-n1.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Because the three alternatives are monotonic, we don't need
to keep a couple of bitmasks, just identify the strongest
alternative at startup.
Generalize test_buffer_is_zero_next_accel and init_accel
by always defining an accel_table array.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Split less-than and greater-than 256 cases.
Use unaligned accesses for head and tail.
Avoid using out-of-bounds pointers in loop boundary conditions.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Increase unroll factor in SIMD loops from 4x to 8x in order to move
their bottlenecks from ALU port contention to load issue rate (two loads
per cycle on popular x86 implementations).
Avoid using out-of-bounds pointers in loop boundary conditions.
Follow SSE2 implementation strategy in the AVX2 variant. Avoid use of
PTEST, which is not profitable there (like in the removed SSE4 variant).
Signed-off-by: Alexander Monakov <amonakov@ispras.ru>
Signed-off-by: Mikhail Romanov <mmromanov@ispras.ru>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240206204809.9859-6-amonakov@ispras.ru>
Use of prefetching in bufferiszero.c is quite questionable:
- prefetches are issued just a few CPU cycles before the corresponding
line would be hit by demand loads;
- they are done for simple access patterns, i.e. where hardware
prefetchers can perform better;
- they compete for load ports in loops that should be limited by load
port throughput rather than ALU throughput.
Signed-off-by: Alexander Monakov <amonakov@ispras.ru>
Signed-off-by: Mikhail Romanov <mmromanov@ispras.ru>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240206204809.9859-5-amonakov@ispras.ru>
Test for length >= 256 inline, where is is often a constant.
Before calling into the accelerated routine, sample three bytes
from the buffer, which handles most non-zero buffers.
Signed-off-by: Alexander Monakov <amonakov@ispras.ru>
Signed-off-by: Mikhail Romanov <mmromanov@ispras.ru>
Message-Id: <20240206204809.9859-3-amonakov@ispras.ru>
[rth: Use __builtin_constant_p; move the indirect call out of line.]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Thanks to early checks in the inline buffer_is_zero wrapper, the SIMD
routines are invoked much more rarely in normal use when most buffers
are non-zero. This makes use of AVX512 unprofitable, as it incurs extra
frequency and voltage transition periods during which the CPU operates
at reduced performance, as described in
https://travisdowns.github.io/blog/2020/01/17/avxfreq1.html
Signed-off-by: Mikhail Romanov <mmromanov@ispras.ru>
Signed-off-by: Alexander Monakov <amonakov@ispras.ru>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240206204809.9859-4-amonakov@ispras.ru>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
The SSE4.1 variant is virtually identical to the SSE2 variant, except
for using 'PTEST+JNZ' in place of 'PCMPEQB+PMOVMSKB+CMP+JNE' for testing
if an SSE register is all zeroes. The PTEST instruction decodes to two
uops, so it can be handled only by the complex decoder, and since
CMP+JNE are macro-fused, both sequences decode to three uops. The uops
comprising the PTEST instruction dispatch to p0 and p5 on Intel CPUs, so
PCMPEQB+PMOVMSKB is comparatively more flexible from dispatch
standpoint.
Hence, the use of PTEST brings no benefit from throughput standpoint.
Its latency is not important, since it feeds only a conditional jump,
which terminates the dependency chain.
I never observed PTEST variants to be faster on real hardware.
Signed-off-by: Alexander Monakov <amonakov@ispras.ru>
Signed-off-by: Mikhail Romanov <mmromanov@ispras.ru>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240206204809.9859-2-amonakov@ispras.ru>
When mechanically moving the @dirty field to AccelCPUState
in commit 9ad49538c7, we neglected cpu->accel is still NULL
when we want to dereference it.
Fixes: 9ad49538c7 ("accel/whpx: Use accel-specific per-vcpu @dirty field")
Reported-by: Volker Rümelin <vr_qemu@t-online.de>
Suggested-by: Volker Rümelin <vr_qemu@t-online.de>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240429091918.27429-2-philmd@linaro.org>
When mechanically moving the @dirty field to AccelCPUState
in commit 79f1926b2d, we neglected cpu->accel is still NULL
when we want to dereference it.
Reported-by: Volker Rümelin <vr_qemu@t-online.de>
Suggested-by: Volker Rümelin <vr_qemu@t-online.de>
Fixes: 79f1926b2d ("accel/nvmm: Use accel-specific per-vcpu @dirty field")
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240429091918.27429-3-philmd@linaro.org>
