[Bug binutils/30725] severe objdump performance regression

2023-08-19 Thread Stromeko at nexgo dot de
https://sourceware.org/bugzilla/show_bug.cgi?id=30725

--- Comment #4 from Achim  ---
Looking at the code in objdump, it relies on bfd to return the line number
information in constant time.  This is what _should_ be happening if the
debuginfo stash is actually working, but apparently it either doesn't for some
files or recreation of info not in the stash takes a really long time so that
O(1) becomes O(N) somehow for the line number lookup.

-- 
You are receiving this mail because:
You are on the CC list for the bug.


[Bug binutils/30725] severe objdump performance regression

2023-08-17 Thread Stromeko at nexgo dot de
https://sourceware.org/bugzilla/show_bug.cgi?id=30725

--- Comment #3 from Achim  ---
No, the debuginfod is not enabled and the option has no effect other than a
warning about E being an unrecognized debug letter option.

Again, both extracting the line number information from the DWARF5 section and
the disassembly alone doesn't produce anthing that looks unusual in terms of
runtime.  Disassembly runtime scales almost lineraly with amount of I/O, DWARF5
record extraction (which must do a fair bit of interpretation of the records I
assume rather seems to scale roughly linearly by number of records.

objdump 2.41-3 -d -l
gcc-ar.exe 0.031u 0.062s  0:00.08 112.5% 0+0k 0+0io  3332pf+0w
gcc-ranlib.exe 0.031u 0.000s  0:00.06  50.0% 0+0k 0+0io  3332pf+0w
gcc-nm.exe 0.046u 0.015s  0:00.07  71.4% 0+0k 0+0io  3344pf+0w
gcov-dump.exe  2.125u 0.046s  0:02.24  96.4% 0+0k 0+0io  4695pf+0w
gcov.exe   2.765u 0.000s  0:02.97  92.9% 0+0k 0+0io  7077pf+0w
collect2.exe   2.953u 0.000s  0:03.26  90.4% 0+0k 0+0io  4746pf+0w
gcov-tool.exe  3.078u 0.031s  0:03.09 100.3% 0+0k 0+0io  4714pf+0w
gnatkr.exe 4.218u 0.030s  0:04.31  98.3% 0+0k 0+0io  8583pf+0w
gnatlink.exe   4.374u 0.000s  0:04.49  97.3% 0+0k 0+0io  9394pf+0w
gnat.exe   4.406u 0.015s  0:04.54  97.1% 0+0k 0+0io  8995pf+0w
lto-wrapper.exe4.718u 0.015s  0:04.96  95.1% 0+0k 0+0io  6682pf+0w
gnatfind.exe   4.859u 0.046s  0:05.09  96.0% 0+0k 0+0io 10604pf+0w
gnatxref.exe   4.984u 0.031s  0:05.32  94.1% 0+0k 0+0io 10596pf+0w
gnatchop.exe   5.125u 0.030s  0:05.22  98.6% 0+0k 0+0io 10846pf+0w
gnatclean.exe  5.828u 0.031s  0:06.14  95.2% 0+0k 0+0io 14015pf+0w
gnatname.exe   5.968u 0.061s  0:06.18  97.4% 0+0k 0+0io 13581pf+0w
gnatprep.exe   6.093u 0.000s  0:06.19  98.3% 0+0k 0+0io 14091pf+0w
gnatls.exe 6.187u 0.000s  0:06.39  96.7% 0+0k 0+0io 14690pf+0w
gnatmake.exe   6.281u 0.031s  0:06.76  93.3% 0+0k 0+0io 15692pf+0w
gfortran.exe   6.593u 0.015s  0:06.98  94.5% 0+0k 0+0io  6736pf+0w
cpp.exe6.703u 0.000s  0:06.97  96.1% 0+0k 0+0io  6731pf+0w
gdc.exe7.000u 0.030s  0:07.00 100.4% 0+0k 0+0io  6735pf+0w
gnatbind.exe  13.406u 0.000s  0:13.88  96.5% 0+0k 0+0io  7892pf+0w
lto-dump.exe1764.452u 7.656s 32:35.54  90.6% 0+0k 0+0io 34002pf+0w
lto1.exe1799.812u 7.406s 33:06.78  90.9% 0+0k 0+0io 34351pf+0w
f951.exe2075.375u 6.562s 37:54.62  91.5% 0+0k 0+0io 35373pf+0w
cc1.exe 2105.890u 7.593s 38:24.45  91.7% 0+0k 0+0io 34919pf+0w
gnat1.exe   2281.406u 6.656s 41:16.20  92.4% 0+0k 0+0io 36665pf+0w
d21.exe 2320.968u 8.031s 42:03.11  92.3% 0+0k 0+0io 36393pf+0w
cc1plus.exe 2322.546u 7.812s 42:01.87  92.4% 0+0k 0+0io 37496pf+0w
cc1objplus.exe  2398.765u 7.859s 43:19.73  92.5% 0+0k 0+0io 37661pf+0w

objdump 2.41-3 -d
gcc-ar.exe 0.015u 0.000s 0:00.03  33.3%  0+0k 0+0io  3388pf+0w 5523
gcc-ranlib.exe 0.015u 0.015s 0:00.03  66.6%  0+0k 0+0io  3387pf+0w 5523
gcc-nm.exe 0.031u 0.015s 0:00.03 133.3%  0+0k 0+0io  3389pf+0w 5523
gcov-dump.exe  0.484u 0.000s 0:00.60  80.0%  0+0k 0+0io  3966pf+0w   143895
collect2.exe   0.546u 0.000s 0:00.68  79.4%  0+0k 0+0io  4012pf+0w   160080
gcov-tool.exe  0.546u 0.000s 0:00.62  87.0%  0+0k 0+0io  3997pf+0w   150923
cpp.exe0.703u 0.015s 0:01.02  69.6%  0+0k 0+0io  4494pf+0w   241130
lto-wrapper.exe0.796u 0.000s 0:00.89  88.7%  0+0k 0+0io  4426pf+0w   218062
gfortran.exe   0.875u 0.062s 0:00.98  94.8%  0+0k 0+0io  4487pf+0w   241676
gcov.exe   0.890u 0.015s 0:00.89 101.1%  0+0k 0+0io  4780pf+0w   218018
gdc.exe0.921u 0.015s 0:00.98  94.8%  0+0k 0+0io  4488pf+0w   241859
gnatkr.exe 0.936u 0.000s 0:01.00  93.0%  0+0k 0+0io  4638pf+0w   244411
gnat.exe   0.953u 0.015s 0:01.03  93.2%  0+0k 0+0io  4694pf+0w   259251
gnatlink.exe   1.015u 0.030s 0:01.11  93.6%  0+0k 0+0io  4759pf+0w   274573
gnatxref.exe   1.187u 0.030s 0:01.30  93.0%  0+0k 0+0io  4866pf+0w   328849
gnatfind.exe   1.218u 0.015s 0:01.27  96.0%  0+0k 0+0io  4871pf+0w   329068
gnatchop.exe   1.437u 0.000s 0:01.33 107.5%  0+0k 0+0io  4894pf+0w   334560
gnatprep.exe   1.625u 0.015s 0:01.83  89.0%  0+0k 0+0io  5393pf+0w   467170
gnatclean.exe  1.703u 0.015s 0:01.82  93.9%  0+0k 0+0io  5389pf+0w   464192
gnatname.exe   1.718u 0.031s 0:01.76  98.8%  0+0k 0+0io  5325pf+0w   447727
gnatls.exe 1.843u 0.000s 0:01.96  93.8%  0+0k 0+0io  5473pf+0w   491418
gnatbind.exe   1.984u 0.015s 0:02.23  89.2%  0+0k 0+0io  5506pf+0w   563078
gnatmake.exe   2.046u 0.015s 0:02.09  98.0%  0+0k 0+0io  5585pf+0w   447727
lto-dump.exe  15.359u 0.015s 0:16.16  95.0%  0+0k 0+0io 30823pf+0w  6836384
lto1.exe  15.390u 0.046s 0:16.41  94.0%  0+0k 0+0io 30811pf+0w  6836218
cc1.exe   16.562u 0.078s 0:17.90  92.9%  0+0k 0+0io 31598pf+0w  7102189
d21.exe   17.265u 0.156s 0:18.38  94.7%  0+0k 0+0io 32801pf+0w  

[Bug binutils/30725] severe objdump performance regression

2023-08-15 Thread nickc at redhat dot com
https://sourceware.org/bugzilla/show_bug.cgi?id=30725

Nick Clifton  changed:

   What|Removed |Added

 CC||nickc at redhat dot com

--- Comment #2 from Nick Clifton  ---
Hi Achim,

  Is the Cygwin version of objdump build with debuginfod support enabled ?

  If so, does adding -wE or --dwarf=do-not-use-debuginfod to the command line
  improve the performance times ?

  I suspect that the presence of the -l option is the trigger for the slow
  down, since this means that the debug information has to be processed in
  order to determine the line numbers.  Processing DWARF is a slow business
  and it has only become slower over successive releases as bug fixing means
  that more care has to be taken to handle all the possible intricacies.

  Hence - as a workaround, could you omit the use of the -l option ?

Cheers
  Nick

-- 
You are receiving this mail because:
You are on the CC list for the bug.


[Bug binutils/30725] severe objdump performance regression

2023-08-06 Thread Stromeko at nexgo dot de
https://sourceware.org/bugzilla/show_bug.cgi?id=30725

--- Comment #1 from Achim  ---
The preformance regression is unrelated to bug#30724.  Results for running
"objdump -l -d" in various versions on the same gcc-11.4 build directory:

objdump 2.39
gcc-ar.exe0.000u 0.000s  0:00.03   0.0%  0+0k 0+0io  3435pf+0w
gcc-nm.exe0.000u 0.000s  0:00.03   0.0%  0+0k 0+0io  3436pf+0w
gcc-ranlib.exe0.015u 0.000s  0:00.04  25.0%  0+0k 0+0io  3439pf+0w
gcov-dump.exe 1.046u 0.000s  0:01.05  99.0%  0+0k 0+0io  4816pf+0w
collect2.exe  1.187u 0.000s  0:01.39  84.8%  0+0k 0+0io  4875pf+0w
gcov-tool.exe 1.250u 0.015s  0:01.37  91.9%  0+0k 0+0io  4857pf+0w
gcov.exe  1.390u 0.000s  0:01.50  92.6%  0+0k 0+0io  7264pf+0w
lto-wrapper.exe   1.796u 0.015s  0:02.05  87.8%  0+0k 0+0io  6865pf+0w
gnatkr.exe2.000u 0.030s  0:02.03 100.0%  0+0k 0+0io  8786pf+0w
gnat.exe  2.062u 0.015s  0:02.17  95.3%  0+0k 0+0io  9225pf+0w
gnatlink.exe  2.093u 0.015s  0:02.24  93.7%  0+0k 0+0io  9614pf+0w
gnatfind.exe  2.436u 0.015s  0:02.55  95.6%  0+0k 0+0io 10834pf+0w
gnatchop.exe  2.453u 0.046s  0:02.59  96.1%  0+0k 0+0io 11102pf+0w
gnatxref.exe  2.561u 0.031s  0:02.51 103.1%  0+0k 0+0io 10843pf+0w
gfortran.exe  2.671u 0.031s  0:02.60 103.8%  0+0k 0+0io  6934pf+0w
gnatclean.exe 3.250u 0.031s  0:03.62  90.6%  0+0k 0+0io 14318pf+0w
gnatprep.exe  3.296u 0.015s  0:03.49  94.5%  0+0k 0+0io 14389pf+0w
gnatname.exe  3.312u 0.015s  0:03.40  97.6%  0+0k 0+0io 13866pf+0w
gnatmake.exe  3.484u 0.046s  0:03.80  92.6%  0+0k 0+0io 16010pf+0w
gnatls.exe3.562u 0.031s  0:03.53 101.6%  0+0k 0+0io 14998pf+0w
gnatbind.exe  5.312u 0.031s  0:05.55  96.2%  0+0k 0+0io  8087pf+0w
lto-dump.exe569.250u 0.750s 10:09.66  93.4%  0+0k 0+0io 35970pf+0w
lto1.exe571.968u 0.390s 10:11.07  93.6%  0+0k 0+0io 35402pf+0w
f951.exe755.796u 0.468s 13:17.20  94.8%  0+0k 0+0io 36542pf+0w
cc1.exe 758.156u 0.515s 13:19.18  94.9%  0+0k 0+0io 36959pf+0w
gnat1.exe   810.328u 0.702s 14:12.46  95.1%  0+0k 0+0io 38941pf+0w
cc1objplus.exe  901.734u 0.812s 15:45.78  95.4%  0+0k 0+0io 39698pf+0w
cc1plus.exe 977.656u 0.578s 16:59.69  95.9%  0+0k 0+0io 39439pf+0w
d21.exe1106.406u 0.421s 19:09.61  96.2%  0+0k 0+0io 38081pf+0w

objdump 2.40
gcc-ar.exe0.000u 0.000s  0:00.03   0.0%  0+0k 0+0io  3458pf+0w
gcc-nm.exe0.000u 0.000s  0:00.03   0.0%  0+0k 0+0io  3449pf+0w
gcc-ranlib.exe0.000u 0.000s  0:00.04   0.0%  0+0k 0+0io  3454pf+0w
gcov-dump.exe 0.937u 0.000s  0:01.12  83.0%  0+0k 0+0io  4846pf+0w
collect2.exe  1.187u 0.000s  0:01.36  86.7%  0+0k 0+0io  4885pf+0w
gcov-tool.exe 1.327u 0.046s  0:01.30 104.6%  0+0k 0+0io  4870pf+0w
gcov.exe  1.406u 0.000s  0:01.50  93.3%  0+0k 0+0io  7256pf+0w
gnat.exe  1.765u 0.031s  0:02.07  86.4%  0+0k 0+0io  9176pf+0w
lto-wrapper.exe   1.781u 0.030s  0:01.84  98.3%  0+0k 0+0io  6836pf+0w
gnatkr.exe1.953u 0.062s  0:01.95 103.0%  0+0k 0+0io  8762pf+0w
gnatlink.exe  2.109u 0.015s  0:02.11 100.0%  0+0k 0+0io  9578pf+0w
cpp.exe   2.218u 0.031s  0:02.53  88.5%  0+0k 0+0io  6916pf+0w
gnatfind.exe  2.296u 0.061s  0:02.50  94.0%  0+0k 0+0io 10787pf+0w
gnatchop.exe  2.358u 0.046s  0:02.57  92.9%  0+0k 0+0io 11017pf+0w
gnatxref.exe  2.359u 0.015s  0:02.45  96.3%  0+0k 0+0io 10763pf+0w
gdc.exe   2.436u 0.015s  0:02.63  92.7%  0+0k 0+0io  6905pf+0w
gfortran.exe  2.453u 0.015s  0:02.56  96.0%  0+0k 0+0io  6907pf+0w
gnatprep.exe  3.093u 0.031s  0:03.35  93.1%  0+0k 0+0io 14292pf+0w
gnatname.exe  3.125u 0.015s  0:03.17  98.7%  0+0k 0+0io 13776pf+0w
gnatclean.exe 3.218u 0.093s  0:03.46  95.3%  0+0k 0+0io 14215pf+0w
gnatmake.exe  3.406u 0.061s  0:03.72  93.0%  0+0k 0+0io 15897pf+0w
gnatls.exe3.484u 0.030s  0:03.52  99.7%  0+0k 0+0io 14897pf+0w
gnatbind.exe  5.250u 0.030s  0:05.27 100.1%  0+0k 0+0io  8061pf+0w
lto1.exe578.750u 0.640s 10:20.99  93.3%  0+0k 0+0io 35910pf+0w
lto-dump.exe586.765u 0.531s 10:30.07  93.2%  0+0k 0+0io 35379pf+0w
f951.exe792.859u 0.500s 13:58.40  94.6%  0+0k 0+0io 36740pf+0w
cc1.exe 842.187u 0.640s 14:50.09  94.6%  0+0k 0+0io 36959pf+0w
gnat1.exe   865.765u 0.734s 15:15.06  94.6%  0+0k 0+0io 39099pf+0w
cc1plus.exe 935.328u 0.531s 16:22.22  95.2%  0+0k 0+0io 39441pf+0w
cc1objplus.exe  962.718u 0.375s 16:48.90  95.4%  0+0k 0+0io 39544pf+0w
d21.exe1116.828u 0.531s 19:24.78  95.9%  0+0k 0+0io 38061pf+0w

objdump 2.41
gcc-ar.exe0.031u 0.000s  0:00.05  60.0%  0+0k 0+0io  3445pf+0w
gcc-nm.exe0.031u 0.000s  0:00.06  50.0%  0+0k 0+0io  3449pf+0w
gcc-ranlib.exe0.031u 0.000s  0:00.04  75.0%  0+0k 0+0io  3461pf+0w
gcov-dump.exe 2.031u 0.046s  0:02.17  95.3%  0+0k 0+0io  4828pf+0w
collect2.exe  2.859u 0.000s  0:03.11  91.6%  0+0k 0+0io  4888pf+0w
gcov.exe  2.906u 0.000s  0:03.06  94.7%  0+0k 0+0io  7215pf+0w
gcov-tool.exe 2.953u 0.015s  0:0

[Bug binutils/30725] severe objdump performance regression

2023-08-05 Thread sam at gentoo dot org
https://sourceware.org/bugzilla/show_bug.cgi?id=30725

Sam James  changed:

   What|Removed |Added

 CC||sam at gentoo dot org

-- 
You are receiving this mail because:
You are on the CC list for the bug.


[Bug binutils/30725] severe objdump performance regression

2023-08-05 Thread ssbssa at sourceware dot org
https://sourceware.org/bugzilla/show_bug.cgi?id=30725

Hannes Domani  changed:

   What|Removed |Added

 CC||ssbssa at sourceware dot org

-- 
You are receiving this mail because:
You are on the CC list for the bug.