reset accumulated profiler samples causes an access violation error

[in-league-with-satan]

starting from commit 8048868 game crashes with exception 0xc0000005

tried self builds and a few artifacts from github-actions. for example this one

Q2RTX_CrashReport01.txt

[in-league-with-satan]

looks like the same error
#293

[res2k]

Unfortunately, I'm struggling to reproduce the error.

Could you either
(a) try the build here?: https://github.com/res2k/Q2RTX/actions/runs/5822525998
It does not fix the error, but should hopefully produce a more expressive crash report.
or,
(b) provide the complete callstack from your own build? (I presume that's why you could tell it's in the profiler sample reset.)

Thanks!

[in-league-with-satan]

https://github.com/res2k/Q2RTX/actions/runs/5822525998 - runs without problems
and my build with pdb start works too

i don't use debugger, just iterated back from master all the commits until it worked on 480f3ec, looks what changed, and then disabled function reset_samples in master and master start works to. for me this way is easier because i can't use IDE and debugging in command line is too hard for me x_x

[in-league-with-satan]

hm...

in reset_samples

we get pointer by idx index
profiler_entry_samples_t *entry_samples = &profiler_data.samples[idx];

and index enumrates in vkpt_profiler_next_frame through
for (int idx = 0; idx < NUM_PROFILER_QUERIES_PER_FRAME; idx++)

#define NUM_PROFILER_QUERIES_PER_FRAME (NUM_PROFILER_ENTRIES * 2)

and size of samples is
profiler_entry_samples_t samples[NUM_PROFILER_ENTRIES];

index out of range of samples size

[apanteleev]

Thanks for the investigation @in-league-with-satan - I fixed the indexing issue in 63cc413