ReadExternalTargetMemory reads bytes out of the target's address space:
process_vm_readv for live bytes (gated like ptrace access), falling back
to the target's on-disk image located by scanning /proc/<pid>/maps
directly - the caller runs while the symbol thread may be rebuilding the
shared image cache on refresh.
The resolver previously captured only the first pcinfo result, so inline
frames were lost, and reported ELF virtual addresses the server could
never re-map to the target. Mirror the in-process pcinfo+syminfo flow,
and report symAddr as the target VMA (loadBias + ELF vaddr) so the
server's re-queries can resolve it.
backtrace_create_state_for_file opens the file lazily on first use and,
on a failed open, silently falls back to /proc/self/exe - so a state
created from a path the monitor cannot open would symbolize the target's
addresses against the monitor's own binary. Probe the target's root path
(and, for unlinked images, the mapping's map_files entry), and only
create a state for a path that actually opens.
The list only ever added entries, so a mapping the target re-mapped or
unloaded stayed in the list and could shadow the new mapping or keep
serving an already-unloaded library. Rebuild it from the target's
current maps on each refresh, carrying over the cached backtrace state
for unchanged mappings and dropping entries the current maps no longer
confirm. The dropped entries' paths are abandoned rather than freed:
queued callstack items still carry the path pointer, which the worker
thread reads at its own pace. Strip the kernel's " (deleted)" suffix so
unlinked images converge to their canonical path.
The minVaddr heuristic assumes the mapping's file offset is measured
against the lowest PT_LOAD vaddr of the image. The kernel maps each
PT_LOAD independently, at load_bias + ELF_PAGESTART(p_vaddr) from
p_offset - ELF_PAGEOFFSET(p_vaddr) (fs/binfmt_elf.c, elf_map), so the
assumption only holds for the segment carrying the minimum vaddr. Match
the mapping to its PT_LOAD through the exact offset relation the kernel
used, and take the base from that segment's own vaddr; the minVaddr
heuristic remains as the fallback. The header must be ELF64
(e_ident[4]): the fixed-size elf_ehdr/elf_phdr structs misparse other
classes, and a false segment match on misparsed headers would return a
silently wrong base.
The image paths from /proc/<pid>/maps are only valid in the target's
mount namespace, so resolve them through /proc/<pid>/root (proc(5)). For
unlinked images the kernel keeps the bytes alive in the mapping's
map_files entry, which is opened directly instead.
The state opens the file lazily on first use (fileline_initialize), so
it must own the filename: the caller's buffer may be freed as soon as
this returns. Pass a copy into backtrace_create_state.
Otherwise on unmap you can get the validation layer error:
"ID3D12Resource3::ID3D12Resource::Unmap: pWrittenRange does not point to an empty D3D12_RANGE and
the heap type is D3D12_HEAP_TYPE_READBACK. Readback resources can be written by the CPU but there's
not much utility. The rationale is that readback heaps are stuck in COPY_DEST state such that the
GPU can never use what the CPU is writing. The range [0, 524288) should be empty (Begin >= End)."
Check also https://learn.microsoft.com/en-us/windows/win32/api/d3d12/nf-d3d12-id3d12resource-unmap
"This indicates the region the CPU might have modified" -> tracy essentially declares the entire range as written, but it's not even a cpu->gpu write buffer -> bug.
SysTraceStart previously reported success even when no perf events
could be opened, leaving a worker thread running over no buffers
and an empty CPU section without explanation; it now fails in that
case. Also log when /proc/kallsyms cannot be read, which otherwise
silently results in ??? kernel stack frames.
perf_event_attr.use_clockid only exists since Linux 4.1 and
sample_max_stack since Linux 4.8; older kernels (e.g. the 3.18
kernels of 32-bit Android devices) reject the attributes with
EINVAL/E2BIG, which silently disabled all of system tracing.
Classify the running kernel from uname(2) and only send the fields
its perf_event_open ABI supports.
The symbol worker thread, which answers source code queries, only exists
when call stack support is compiled in. Handle the query directly in the
main thread so it is always answered and the server can terminate.
The mrc to c13/c0/3 (thread id register) is undefined on ARMv5 (e.g.
ARM926EJ-S) and SIGILL'd during init. __builtin_thread_pointer lowers
to the same mrc on targets with a thread pointer register and to
__aeabi_read_tp on soft-TLS targets.
Delayed init was introduced 2019-02-19 in ef5e30056 and was lazy init at
that time, matching the name.
This changed 2020-05-19 in 4eb78f5c8, when auto-init was added, making
delayed init not delayed anymore.
While the option is still needed to make manual lifetime work, and it has
to be enabled on apple, there's no reason anymore to expose it to users.
The macro fast paths only reference GetProfiler from -O0 code, direct
calls, and ON_DEMAND builds: the single GetProfiler() call on the macro
path sits behind the constexpr-dead callstack guard and is folded away
at -O1+, so Release builds carried no config fingerprint and mismatched
clients linked silently.
GetToken is the hot-path binding of every queueing macro and is
referenced at all optimization levels; mangle it the same way so
mismatches fail at link time in Release builds too.
The gpu backends each used a bare 255 as the sentinel for a
not-yet-initialized context id. Define InvalidGpuContextId (-1) in
TracyQueue.hpp and replace the scattered 255s and their asserts
with it.
Id exhaustion is handled by an error message and an assert in the
new NextGpuContextId() getter. Non-assert builds continue execution,
at which point they are no longer valid. Handling this code path
is out of scope here. Various attempts at handling the exhaustion
problem have been otherwise purged from the API implementations.
The decoder recovered offset-encoded 16-bit string lengths into a
uint16_t, so sz += ProtocolOffset8Bit truncated lengths in [65536,
65791] back into [0, 255] in release builds, desynchronizing the
stream. Read the wire value into uint16_t sz16 and recover into a
uint32_t sz, relying on automatic promotion for the addition.
Align the client asserts with the encoder's actual capacity
(ProtocolOffset8Bit + uint16 max).
CUDACtx's constructor writes GpuNewContext directly through
QueueSerialFinish(), unlike every other GPU backend (Vulkan, OpenGL,
D3D11/12, Metal, WebGPU, Rocprof), which all defer it via
GetProfiler().DeferItem() so it survives on-demand's per-connection
queue clear.
A profiler connecting any time after the CUDA context is created (in
practice: any time after process start) never receives GpuNewContext.
The GpuContextName message that Name() sends right after (already
correctly deferred) then crashes the server's
Worker::ProcessGpuContextName with an unregistered context id
(assert(ctx) fails; undefined behavior in release builds).
Same fix already applied to the Rocprof backend in #1336. Fixes#1171.
Includes a repro test under tests/cuda/repro/on_demand/, mirroring the
structure #1336 added for Rocprof: a minimal CUDA program that creates
an on-demand context (repro.cu/CMakeLists.txt), and a check_gpu_zones
tool that loads the resulting .tracy file and verifies the GPU context
was named and populated with zones. Verified locally: unpatched
tracy-capture crashes on the first connection attempt; patched, three
consecutive connect/disconnect cycles all succeed and check_gpu_zones
reports a named context with recorded zones.
