Some programs like RTOS firmware may have a large number of symbols.
The profile information in the profile data file includes histogram
records, which capture low PC and high PC of program execution. If all
histogram records come in the profile data file before any call-graph
records and basic-block records, we can look up only the line numbers
within low PC and high PC in histogram records, which reduces processing
time for such a firmware from ~2 minutes to ~2 seconds.
Add symbol table access function, get_symtab, get_symtab_direct and
set_symtab to delay loading the symbol table until its first use.
* aarch64.c (aarch64_find_call): Call get_symtab to get the
symbol table pointer
* alpha.c (alpha_find_call): Likewise.
* basic_blocks.c (bb_read_rec): Likewise.
(bb_write_blocks): Likewise.
(print_exec_counts): Likewise.
(print_annotated_source): Likewise.
* call_graph.c (cg_tally): Likewise.
(cg_write_arcs): Likewise.
* cg_arcs.c (cycle_link): Likewise.
(propagate_flags): Likewise.
(cg_assemble): Likewise.
* cg_print.c (cg_print): Likewise.
(cg_print_index): Likewise.
(cg_print_function_ordering): Likewise.
* corefile.c: Include "gmon_io.h".
(core_create_syms_from): Call get_symtab_direct to get the
symbol table pointer.
(core_create_function_syms): Likewise.
(core_create_line_syms): Likewise. If all histogram records
come in the profile data file before any call-graph records and
basic-block records, we can look up only the line numbers within
low PC and high PC in histogram records.
* gmon_io.c (gmon_histograms_first): New.
(gmon_out_read): Set gmon_histograms_first to true if all
histogram records come first.
(gmon_out_write): Call get_symtab to get the symbol table
pointer.
* hist.c (scale_and_align_entries): Likewise.
(hist_assign_samples_1): Likewise.
(hist_print): Likewise.
* i386.c (i386_find_call): Likewise.
* mips.c (mips_find_call): Likewise.
* sparc.c (sparc_find_call): Likewise.
* sym_ids.c (sym_id_parse): Likewise.
* vax.c (vax_find_call): Likewise.
* gmon_io.h (gmon_histograms_first): New.
* gprof.c (man): Don't create profile info.
(symtab_init): New.
* gprof.h (symtab_init): New.
* symtab.c (symtab): Changed to static.
(get_symtab_direct): New.
(get_symtab): Likewise.
(set_symtab): Likewise.
* symtab.h (symtab): Removed.
(get_symtab_direct): New.
(get_symtab): Likewise.
(set_symtab): Likewise.
Signed-off-by: Richard Allen <rsaxvc@gmail.com>
Co-Authored-By: H.J. Lu <hjl.tools@gmail.com>
With even instruction sizes, this rounding never truncated.
Xtensa CPUs mostly use 2-3 byte instructions, and this can lead
to a histogram being captured with an odd length address range.
This small truncation prevented gprof from parsing gmon.out files
containing multiple histograms when at least one of them has an
odd address range length and another has any other address range.
Signed-off-by: Richard Allen <rsaxvc@gmail.com>
Adds two new external authors to etc/update-copyright.py to cover
bfd/ax_tls.m4, and adds gprofng to dirs handled automatically, then
updates copyright messages as follows:
1) Update cgen/utils.scm emitted copyrights.
2) Run "etc/update-copyright.py --this-year" with an extra external
author I haven't committed, 'Kalray SA.', to cover gas testsuite
files (which should have their copyright message removed).
3) Build with --enable-maintainer-mode --enable-cgen-maint=yes.
4) Check out */po/*.pot which we don't update frequently.
The newer update-copyright.py fixes file encoding too, removing cr/lf
on binutils/bfdtest2.c and ld/testsuite/ld-cygwin/exe-export.exp, and
embedded cr in binutils/testsuite/binutils-all/ar.exp string match.
The result of running etc/update-copyright.py --this-year, fixing all
the files whose mode is changed by the script, plus a build with
--enable-maintainer-mode --enable-cgen-maint=yes, then checking
out */po/*.pot which we don't update frequently.
The copy of cgen was with commit d1dd5fcc38ead reverted as that commit
breaks building of bfp opcodes files.
* readelf.c: Remove duplicate declaration of variable do_wide.
* dwarf.h: Add export of do_wide.
* hist.h: Move declaration of histograms and num_histograms
variables from here to...
* hist.c: ...here.
(histograms, num_histograms): New.
* hist.c (find_histogram, find_histogram_for_pc)
(read_histogram_header): New.
(s_lowpc, s_highpc, lowpc, highpc, hist_num_bins)
(hist_sample): Remove.
(hist_read_rec): Use the above, and handle multiple
histogram records with disjoint address ranges.
(hist_write_hist): Support several histogram records.
(scale_and_align_entries): Adjust for multiple histograms.
(hist_assign_samples_1): New.
(hist_assign_samples): Use the above.
(hist_clip_symbol_address): New.
* hist.h (hist_check_address)
(hist_clip_symbol_address): Declare.
* gmon_io.c (gmon_out_read, gmon_out_write): Adjust handling
of legacy format for multiple histogram changes.
* corefile.c (find_call): Check for core_text_space and
clip symbol address range here.
* vax.c (vax_find_call): Don't check for
core_text_space, or clip the symbol's address range here.
Use hist_check_address to check call's target address.
* sparc.c: Likewise.
* tahoe.c: Likewise.
* i386.c: Likewise.
* mips.c: Likewise. Also use core_text_sect->vma as the base
address for code accesses, just like other machine-specific
routines do.
* gprof.texi: Adjust for the new logic.
