Set two breakpoints four bytes apart, both logging, run into the first, then
press Next: the second one never logs, however many times you step. Reproduced
on both the interpreter and the JIT.
The skip-first mechanism was doing two different jobs with one marker. Every
resume and every step recorded the address it started from, and any breakpoint
check at that address was suppressed outright. That's right for the breakpoint
you're parked on - you have to be able to get off it - but stepping *onto* an
address is not the same as having reported the breakpoint there, and the next
step suppressed it before it ever logged.
Split into the two things that were being conflated:
- resumedFrom_ is where the current run or step started. It only drops the
pause, not the log or the hit count. It still covers the temporary breakpoint,
which is what makes "run to here" work when you're already on that address.
- reported_ is the breakpoint we already logged and counted. Reporting stops the
CPU before the instruction runs, so the resume that follows arrives at the
same pending execution and must not report it twice.
Both are (address, tick count) pairs, which identify one pending execution of
one instruction: ticks only move when the CPU retires an instruction, so the
marker stops matching as soon as it runs, and a breakpoint in a loop still fires
every iteration.
reported_ can't be armed where the report happens, though. Under a JIT that's
inside a compiled block whose cycles are already accounted for, so the tick
count there isn't the settled one we see on the way back in - arming it there
double-logged the breakpoint under -j. So the report just records the address,
and NotifyResumingFrom() turns it into a real marker once the CPU has stopped.
That also has to be idempotent: a step-over arms its temporary breakpoint and
then calls Core_Resume(), which notifies a second time.
MemCheck::Action() no longer pauses by itself - the caller decides, the same way
ExecBreakPoint() already did, so all three breakpoint kinds share the handling.
Verified on both backends: two adjacent breakpoints now log once each while
stepping (was one log total), stepping off a breakpoint still doesn't re-log it
(was two under -j), step-over still skips the call and logs a breakpoint at the
address it lands on, and a breakpoint in a loop reports once per iteration.
pspautotests 314/314.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GZq8ZtJmFY7bkX5FVkr3P9
The tick basis was actually sound - CoreTiming::GetTicks() is continuous across
Advance(), so "ticks unchanged" really does mean "no instruction retired since",
which is exactly the window the suppression needs. The plumbing around it was
the problem:
- ExecRegBreakpoint() applied the skip only to the pause, so stepping off a
log+pause register breakpoint printed it again and counted a second hit. The
check now sits at the top of ExecBreakPoint(), ExecMemCheck(), ExecOpMemCheck()
and ExecRegBreakpoint() instead of being repeated at seven call sites across
the interpreter and four JIT frontends, where one of them had it wrong and
another checked a different address than the rest.
- Address 0 doubled as "nothing to skip" (ClearSkipFirst() existed but was dead
code; the JITs cleared by calling SetSkipFirst(0)), so a breakpoint at 0 would
have been permanently suppressed. There's an explicit valid flag now, and
ClearSkipFirst() is what clears it.
- The marker was set from five places and never cleared when execution stopped,
so one could outlive the resume that armed it. Core_Break() clears it now, and
the two WebSocket subscribers that set it immediately before asking for a
step - which sets it again itself - no longer do.
- SetSkipFirst() now only arms when some breakpoint machinery actually exists,
so a stale marker can't sit around waiting to swallow a breakpoint added later.
CheckSkipFirst() returning an address (compared against pc by each caller) is
replaced by ShouldSkipBreakpoint(addr), which compares against both addr and
currentMIPS->pc - under a JIT those differ, and only some callers knew that.
Covered by the Breakpoints unit test, including that a suppressed breakpoint
neither logs nor counts a hit.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GZq8ZtJmFY7bkX5FVkr3P9
step-over, step-out and run-until plant a one-shot breakpoint at the address
they want execution to return to. Keeping it in breakPoints_ alongside the
user's own meant the two kept colliding:
- Adding a log-only user breakpoint at the same address hijacked the temporary
one. AddBreakPoint() didn't match across temp-ness so both existed, and then
ChangeBreakPoint() looked up "the first enabled breakpoint at this address" -
a log-only breakpoint isn't enabled, so the temporary one won and had its
action overwritten to log-only. It lost PAUSE and the step never came back.
- RemoveBreakPoint() erased up to two entries per address to catch an
overlapping temporary one, so deleting either deleted both - including the
interpreter's cleanup path in CheckExecBreakpoints() taking the user's
breakpoint with it.
- ExecBreakPoint() handled one breakpoint per address, so with both at the same
address only one of them did anything: the step completed but the user's log
line never printed.
- Nothing dropped it when something *else* stopped us first, so an interrupted
step left a breakpoint armed at an address nobody was waiting for anymore,
which later fired as a phantom stop.
It's a single TempBreakPoint member now, invisible to the breakpoint lists and
untouched by user edits. One is enough: step over/out and cross-thread step into
all require the CPU to already be stepping and resume it immediately, so only
one can be in flight, and run-until now replaces rather than stacking (two
pending run-untils had no coherent meaning, and the loser stayed armed).
Behavior follows what other debuggers do. Both breakpoints at an address are
evaluated independently and their actions combine, so a log-only breakpoint
logs without stopping and still lets the step finish. Core_Break() drops the
temporary breakpoint on any stop, whatever the reason - the same way gdb deletes
its step-resume breakpoint and lldb discards the thread plan.
Two things to be careful of, both covered by the new TempBreakpoints test:
HasBreakPoints() has to account for it, or the interpreter's checked run loop
and the JIT skip breakpoint checking entirely and a step with no user
breakpoints set never returns; and IsAddressBreakPoint() (user-facing, for the
lists and disassembly markers) is now separate from NeedsBreakCheckAt() (what
the JIT frontends and interpreter ask), since only the latter should see it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GZq8ZtJmFY7bkX5FVkr3P9
All ~82 MIPSInt::Int_* functions (Interpreter.cpp/.h,
InterpreterVFPU.cpp/.h) now take an explicit MIPSState *mips instead
of reaching for the global currentMIPS internally, along with their
file-local helpers (DelayBranchTo, SkipLikely, ApplySwizzleS/T,
ApplyPrefixD/ST, RetainInvalidSwizzleST, EatPrefixes). MIPSInterpretFunc,
Interpret(), ExecInstruction()/InterpreterDispatch.cpp (regenerated),
and RunUntilFast() all thread mips through accordingly.
Deliberately left on currentMIPS for now: MIPSVFPUUtils.cpp's
ReadVector/WriteVector/ReadMatrix/WriteMatrix/VFPURewritePrefix -
these are shared with every JIT backend's compile-time VFPU code, so
parameterizing them would balloon this into a JIT-wide refactor. This
is a partial refactor; that's the next boundary to push on.
Several JIT backends (x86 Jit.cpp, ARM/ArmJit.cpp, ARM64/Arm64Jit.cpp,
x86/X64IRJit.cpp, RiscV/RiscVJit.cpp, LoongArch64/LoongArch64Jit.cpp,
ARM64/Arm64IRJit.cpp) bake the raw interpreter function pointer
directly into JIT-generated machine code as their "fall back to the
interpreter for this one op" mechanism, with only a single argument
register set up for the call. Rather than hand-editing register
allocation across four architectures that can't be build-tested here,
added MIPSInterpretTrampoline(MIPSOpcode op) - a 1-arg wrapper around
MIPSInterpret(currentMIPS, op) - and pointed all 7 such call sites at
it instead, leaving that codegen untouched. Two other call sites
(JitLogMiss, JitBranchLog) were plain C++ calls and just got the
extra argument directly.
Verified (Windows x64): PPSSPPWindows/PPSSPPHeadless/UnitTest all
build clean, 49/49 unit tests pass. `test.py -g --graphics=software`:
interpreter 312/314 (cpu/fpu/fpu is the pre-existing, unrelated
interpreter-vs-JIT denormal difference; gpu/rendertarget/copy passes
standalone, so was cross-test state bleed in the batch run, not a
regression), default JIT 314/314, jit-ir 313/314 (gpu/vertices/morph
is an expected difference from the vertex decoder taking a different
mode with this core change, not a bug).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019SKhm9wKEQzRUsx9mTrrtQ
* Rename LogType to Log
* Explicitly use the Log:: enum when logging. Allows for autocomplete when editing.
* Mac/ARM64 buildfix
* Do the same with the hle result log macros
* Rename the log names to mixed case while at it.
* iOS buildfix
* Qt buildfix attempt, ARM32 buildfix
If a memory breakpoint hits within a replacement (like memcpy, memset) we
would previously move PC back to the jal in cases, which would break
things if you tried to resume.
Otherwise, the block will be executed with the wrong rounding mode the
first time rounding is set. This could be important if it was set for a
single operation.
This is only a problem the first time it's set.