Improve DsoHdr::erase_range

[nsavoire]

• Add DDPROF_CHECK_FATAL / DDPROF_DCHECK_FATAL assertions
• Add function to check DsoHdr invariants
• Change pid_fork to do nothing if child pid mappings already exist
  pid_fork directly copies mappings from parent to child pid,
  if child pid already has mappoings this can lead to overlapping mappings.
  Currently this cannot happens, because child mappings are cleared before
  pid_fork. This check adds a safegard by returning early if child pid
  has mappings.
• Add DsoHdr invariant check in debug mode after inserting a Dso
• Make erase_range function cleaner
  • Add check when re-inserting truncated last mapping
  • Return early in the case new mapping truncates end of a single mapping
    to avoid using range.first ierator that is invalidated by map::extract
    (the node pointed by first is extracted and then re-inserted hence
    the iterator should remain valid, but better be careful)
• Allow anonymous mapping to truncate existing file mapping
  Allow new mapping to truncate existing mappings if previous mapping is
  a file and new mapping is anonymous because a LOAD segment can have
  filesize < memsize as a size optimization when the end of segment is all
  zeros (eg. bss). In that case elf loader / dlopen will map the whole
  file + some extra space to cover the difference between memsize and file
  size and then map an anonymous mapping over the part not in the file.
• Avoid out-of-bound array access when printing unwinding error in debug logs

[pr-commenter]

Benchmark results for collatz

Parameters

	Baseline	Candidate
config	baseline	candidate
profiler-version	ddprof 0.15.3+4e33d7a4.27091101	ddprof 0.15.3+284e23be.27241456

Summary

Found 0 performance improvements and 0 performance regressions! Performance is the same for 1 metrics, 0 unstable metrics.

See unchanged results

scenario	Δ mean execution_time
scenario:ddprof -S bench-collatz --preset cpu_only collatz_runner.sh	same

[pr-commenter]

Benchmark results for BadBoggleSolver_run

Parameters

	Baseline	Candidate
config	baseline	candidate
profiler-version	ddprof 0.15.3+4e33d7a4.27091101	ddprof 0.15.3+284e23be.27241456

Summary

Found 0 performance improvements and 0 performance regressions! Performance is the same for 1 metrics, 0 unstable metrics.

See unchanged results

scenario	Δ mean execution_time
scenario:ddprof -S bench-bad-boggle-solver BadBoggleSolver_run work 1000	same

[r1viollet]

LGTM
We had to do a dedicated session to review some of the changes here.
The erase_range logics are not easy to understand and require a careful understanding of linker behaviours.