464689 - DBX failure in memory leak testing not reported.

Status: RESOLVED DUPLICATE of bug 464223

(NSS :: Test, defect)

Description

[Slavomir Katuscak]

Occured on: aquash SunOS/i386 32bit DBG 2008/11/12

Log:
---
chains.sh: Verifying certificate(s)  EE2CA2.der CA2Bridge.der BridgeArmy.der with flags -d EE1DB -f   -t Navy.der
vfychain -d EE1DB -pp -vv -f   EE2CA2.der CA2Bridge.der BridgeArmy.der  -t Navy.der
chains.sh: -------- Running /export/tinderlight/data/aquash_32_DBG/mozilla/dist/SunOS5.9_i86pc_DBG.OBJ/bin/vfychain under DBX:
/usr/dist/pkgs/sunstudio_i386,v11.0/SUNWspro/prod/bin/dbx /export/tinderlight/data/aquash_32_DBG/mozilla/dist/SunOS5.9_i86pc_DBG.OBJ/bin/vfychain
chains.sh: -------- DBX commands:
dbxenv follow_fork_mode parent
dbxenv rtc_mel_at_exit verbose
dbxenv rtc_biu_at_exit verbose
check -memuse -match 16 -frames 16
run -d EE1DB -pp -vv -f EE2CA2.der CA2Bridge.der BridgeArmy.der -t Navy.der

chains.sh: #2866: BridgeWithHalfAIA: Verifying certificate(s)  EE2CA2.der CA2Bridge.der BridgeArmy.der with flags -d EE1DB -f   -t Navy.der - FAILED
---

When I was trying to get some more data from special logfile for memory leaks, I found out that there are no data from this tests (no output from dbx) - so it looks that DBX failed (can be cause by temporary network problems).

Explanation is in this part of memleak.sh code:
echo "${DBX_CMD}" | ${DBX} ${COMMAND} 2>/dev/null | grep -v Reading > ${TMP_DBX}
cat ${TMP_DBX} 1>&2

If there is any output, than it's all logged. If dbx fails, than there is no output (dbx error messages goes to /dev/null).

For now this is one time problem, if it repeats I will add comments there.

Comment 1

[Slavomir Katuscak]

Have second occurence now:
ciotat SunOS/sparc 64bit OPT on 2008/11/13 03:06:19 (Tinderbox again)

chains.sh: Verifying certificate(s)  EE1CA13.der RootCA.der CA1RootCA.der CA12CA1.der CA13CA12.der with flags -d AllDB   -o OID.2.0  -t RootCA.der
vfychain -d AllDB -pp -vv   -o OID.2.0  EE1CA13.der RootCA.der CA1RootCA.der CA12CA1.der CA13CA12.der  -t RootCA.der
chains.sh: -------- Running /export/tinderlight/data/ciotat_64_OPT/mozilla/dist/SunOS5.9_64_OPT.OBJ/bin/vfychain under DBX:
/usr/dist/pkgs/sunstudio_sparc,v12.0/SUNWspro/prod/bin/dbx /export/tinderlight/data/ciotat_64_OPT/mozilla/dist/SunOS5.9_64_OPT.OBJ/bin/vfychain
chains.sh: -------- DBX commands:
dbxenv follow_fork_mode parent
dbxenv rtc_mel_at_exit verbose
dbxenv rtc_biu_at_exit verbose
check -memuse -match 16 -frames 16
run -d AllDB -pp -vv -o OID.2.0 EE1CA13.der RootCA.der CA1RootCA.der CA12CA1.der CA13CA12.der -t RootCA.der

chains.sh: #942: AnyPolicyWithLevel: Verifying certificate(s)  EE1CA13.der RootCA.der CA1RootCA.der CA12CA1.der CA13CA12.der with flags -d AllDB   -o OID.2.0  -t RootCA.der - FAILED

I'm going to prepare patch for better logging of those failures.

Comment 2

[Slavomir Katuscak]

Attachment: Patch to report dbx errors.

DBX standard output goes to log $TMP_DBX and error output to $TMP_DBXERR.
We don't want to mix those data. Standard output ($TMP_DBX) contains memory leak stacks to parse, those stacks goes to separate log files. Error output ($TMP_DBXERR) now goes to main log file (before was ignored).

Comment 3

[Christophe Ravel]

Comment on attachment 347977 [details] [diff] [review]
Patch to report dbx errors.

r+=christophe

Comment 4

[Slavomir Katuscak]

Attachment: Patch v2.

Patch fixing return value problem (introduced in previous patch).

Comment 5

[Alexei Volkov]

(In reply to comment #2)
> Created an attachment (id=347977) [details]
> Patch to report dbx errors.
This patch has introduced the a regression to memleak testing. All negative test were reported as failed. Patch 348062 resolved the problem. r=alexei

Comment 6

[Slavomir Katuscak]

See 464223 comment 15.