[openssl-dev] "openssl s_client" memory leak
Gisle Vanem
gvanem at yahoo.no
Tue Jan 19 02:25:52 UTC 2016
There seems to be some memory leak(s) somewhere during the life
of 's_client_main()'. Example from a client.bat here:
set OPENSSL_DEBUG_MEMORY=on
echo -e GET /ssltest/viewMyClient.html\r\n | openssl.exe s_client ^
-connect www.ssllabs.com:443
('echo -e' is from Cygwin32).
All .html output gets produced correctly; ends with:
</html>read:errno=0
and dumped okay in Lynx.
But the mem-tracker reports:
[12:06:07] 6930 file=crypto/bio/bio_lib.c, line=68, thread=1092, number=72, address=0x284ef78
72 bytes leaked in 1 chunks
And adding other options show additional leaks.
E.g. '-nbio_test -reconnect' results in 12 leaks:
[12:15:17] 13704 file=crypto/bio/bio_lib.c, line=68, thread=4248, number=72, address=0x2ccf468
[12:15:15] 13016 file=crypto/bio/bio_lib.c, line=68, thread=4248, number=72, address=0x2ccedb0
[12:15:18] 14345 file=crypto/bio/bio_lib.c, line=68, thread=4248, number=72, address=0x2ccf238
[12:15:14] 6906 file=crypto/bio/bio_lib.c, line=68, thread=4248, number=72, address=0x2c7e618
[12:15:17] 13705 file=crypto/bio/bio_lib.c, line=68, thread=4248, number=72, address=0x2ccf2a8
[12:15:15] 13017 file=crypto/bio/bio_lib.c, line=68, thread=4248, number=72, address=0x2ccee20
[12:15:16] 13363 file=crypto/bio/bio_lib.c, line=68, thread=4248, number=72, address=0x2ccf078
[12:15:14] 6907 file=crypto/bio/bio_lib.c, line=68, thread=4248, number=72, address=0x2c7e688
[12:15:18] 14344 file=crypto/bio/bio_lib.c, line=68, thread=4248, number=72, address=0x2ccf0e8
[12:15:17] 14022 file=crypto/bio/bio_lib.c, line=68, thread=4248, number=72, address=0x2cceeb8
[12:15:16] 13364 file=crypto/bio/bio_lib.c, line=68, thread=4248, number=72, address=0x2ccef28
[12:15:17] 14021 file=crypto/bio/bio_lib.c, line=68, thread=4248, number=72, address=0x2ccf628
864 bytes leaked in 12 chunks
line 68 in bio_lib.c is inside BIO_new() which gives me no clue as to
the caller(s). And these order-values in column 2, what exactly are they?
All this is with a ver 0x10100003L, MingW + gcc 5.1 build on Win-10.
Maybe someone can verify this happening on other OSes too?
BTW. 'memleaktest.exe freeit' reports nothing. So the mem-dbg seems okay
in this memtleaktest program. But maybe not in the 'big-picture'?
--
--gv
More information about the openssl-dev
mailing list