Re: [ast-developers] Need fix for memoryleak in ksh

[Michal Hlavinka]

Yes, I wrote that the reproducer he referenced is fixed in latest version, but not all reported memory leaks were fixed and as I've tested it, some were fixed and reappeared later.

See reproducer: test48.sh it tests several leaks.
The reproducers are in the
if [ "$1" = foo ];
sections on top. The rest is just loop and memory meter.



See attached reproducers:

test53.sh - builtins -d foo must be called several times







test71.sh - unsetting running function from another one crashes ksh

test73.sh - wrong return code after pipe in function

test75.sh - wrong return code after pipe in ``

When I run tests from the sources, following tests fail:

test namespace(shcomp) begins at 2014-08-19+16:29:42

discipline functions for types in namespace not working



test signal begins at 2014-08-19+16:31:16
./shtests[333]: eval: line 1: 26930: Memory fault(coredump)

test treemove(shcomp) begins at 2014-08-19+16:36:25





Let me know if you need more information to any of those reproducers

On 08/18/2014 11:29 PM, David Korn wrote:

Also if you plan to release stable version, there are still some issues remaining that I've reported some time ago. test54.sh - ksh hangs when pipe buffer is full, can pass on some systems that has wmem_default set to high values ( sysctl -w net.core.wmem_default=65536 ) test56.sh - substitution waits for background process, this is change in behaviour compared to 2012-08-01 (and older), also probably another incompatibility with fake bash. I'm not sure whether this change was really intentional and whether you want to keep it that way. test57.sh - reading file through command substitution does not work when stdout and/or stderr is closed /tmp/tmpgX9z6DF.0By/shcomp-namespace.ksh[127]: .a.b.x_t: not found [No such file or directory] /tmp/tmpgX9z6DF.0By/shcomp-namespace.ksh[128]: var.pi: not found [No such file or directory] /tmp/tmpgX9z6DF.0By/shcomp-namespace.ksh[138]: .com.foo.test1.y_t: not found [No such file or directory] /tmp/tmpgX9z6DF.0By/shcomp-namespace.ksh[139]: v.x.pr: not found [No such file or directory] shcomp-namespace.ksh[139]: _.__ not working with nested types in a namespace /home/mhlavink/gitf/kshbeta/ksh-20140625/arch/linux.i386-64/bin/ksh: line 2: syntax error at line 6: `function' unexpected [78]: typeset -C c=(objstack_t ost=(typeset -l -i st_n=1;st[0]=(obj=(typeset -l -i val=5)))) is not idempotent Also there is one for which I have no reproducer and no backtrace. Since last year, when I try to compile latest beta release, it fails because ksh crashes all the time.

I am not showing a leak with 2014-06-25.  Here is the output with the
middle lines omitted.
I hope to have an official release this fall so send any problems ASAP.
However, some new features (such as programmable completion and json
input are likely to have errors which will be resolved with a patch.



Version ABIJM 93v- 2014-06-25
total kB           15648    2076     368
0.2324 total kB           15648    2192     396
0.2289 total kB           15648    2192     396
0.2278 total kB           15648    2192     396
....
0.2273 total kB           15648    2196     400
0.2261 total kB           15648    2196     400
0.2263 total kB           15648    2196     400
0.3453 total kB           15648    2196     400
0.2953 total kB           15648    2196     400



On Mon, Aug 18, 2014 at 6:47 AM, Michal Hlavinka <mhlav...@redhat.com
<mailto:mhlav...@redhat.com>> wrote:

    On 08/18/2014 08:16 AM, K Sujatha wrote:

        Hi,
              WIth reference to the post,
        
http://lists.research.att.com/__pipermail/ast-developers/__2014q1/003815.html
        
<http://lists.research.att.com/pipermail/ast-developers/2014q1/003815.html>
        is there  a  ksh version with  fix for the memory leak issue.
        I would  be   grateful  for any help.

        Thanks & Regards,
        Sujatha


    Hi,

    as far as I know, latest beta realeases should have the referenced
    memory leak fixed, but there is no version that would fix all known
    memory leaks.

    Michal
    _________________________________________________
    ast-developers mailing list
    ast-developers@lists.research.__att.com
    <mailto:ast-developers@lists.research.att.com>
    http://lists.research.att.com/__mailman/listinfo/ast-__developers
    <http://lists.research.att.com/mailman/listinfo/ast-developers>

test48.sh
Description: application/shellscript

test53.sh
Description: application/shellscript

test54.sh
Description: application/shellscript

test56.sh
Description: application/shellscript

test57.sh
Description: application/shellscript

test71.sh
Description: application/shellscript

test73.sh
Description: application/shellscript

test75.sh
Description: application/shellscript

_______________________________________________
ast-developers mailing list
ast-developers@lists.research.att.com
http://lists.research.att.com/mailman/listinfo/ast-developers