On Tue, 2018-08-28 at 01:45 -0500, Sasha Toltec wrote: > Here are the valgrind errors that were requested. >
Hi Sasha, Thank you for your prompt response. > The problem seems to be that if you don't understand what types of > numbers put stress on a bignum library, then it is > difficult/impossible to force the memory errors. In your particular example below, the case "sqrt(.1)" would have worked just as well; the full `.0000000000123` is unnecessary. > If one of the senior programmers involved with toybox has time to take > a look at the bc bignum library and suggest some fixes and reading > material to its authors that would perhaps be the best course of > action. Toltec Enterprises generally charges for consultation ergo > cannot offer much more input. It would be unwise to retain the consulting services of a group that both (a) denigrates the developers of a community project and (b) neglects to compile a project with the '-g' flag before posting an otherwise useless Valgrind report. [1] and [2]: > sasha@sasha:~/toybox$ uname -a > Linux sasha 4.4.0-103-generic #126-Ubuntu SMP Mon Dec 4 16:23:28 UTC > 2017 x86_64 x86_64 x86_64 GNU/Linux > sasha@sasha:~/toybox$ make bc > sasha@sasha:~/toybox$ echo "sqrt(.0000000000123)" | valgrind ./bc > ==10636== Memcheck, a memory error detector > ==10636== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al. > ==10636== Using Valgrind-3.11.0 and LibVEX; rerun with -h for copyright info > ==10636== Command: ./bc -l > ==10636== > ==10636== Invalid read of size 1 > ==10636== at 0x402D68: ??? (in /home/sasha/toybox/bc) > ==10636== by 0x403C3B: ??? (in /home/sasha/toybox/bc) > ==10636== by 0xFFEFFF84F: ??? > ==10636== by 0x15: ??? > ==10636== by 0x5945C9F: ??? > ==10636== by 0x15: ??? > ==10636== by 0x16: ??? > ==10636== by 0x23: ??? > ==10636== Address 0x5945aef is 1 bytes before a block of size 55 alloc'd > ==10636== at 0x4C2DB8F: malloc (in > /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) > ==10636== by 0x402FFB: ??? (in /home/sasha/toybox/bc) > ... > If you have particular issues with the `bc' project, maybe it would be better to address those with the `bc' authors? This mailing list may not be the appropriate venue for technical discussion of a particular memory issue in one of the "pending" toys (whereas integration and conformance issues would). This `bc' is, in itself, dozens to hundreds of times larger than the other toys. Zach van Rijn [1]: http://lists.landley.net/pipermail/toybox-landley.net /2018-August/009634.html and previous emails from "Sasha". [2]: http://lists.landley.net/pipermail/toybox-landley.net /2018-August/009635.html _______________________________________________ Toybox mailing list Toybox@lists.landley.net http://lists.landley.net/listinfo.cgi/toybox-landley.net