Oddity in scan-build output

Clang Frontend
#1

Hi,

I've noticed that occasionally the "x diagnostics generated" output is mixed into "ANALYZE: ..." lines in scan-build output, as in the following snippet:

ANALYZE: /home/s/projects/dailybuild/src/sys/dev/sound/pcm/sound.c pcm_chn_remove
ANALYZE: /home/s/projects/dailybuild/src/sys/dev/sound/pcm/sound.c pcm_addchan
ANALYZE: /home/s/projects/dailybuild/src/sys/dev/sound/pcm/sound.c pcm_killchan
ANALYZE: /home/s/projects/dailybuild/src/sys/dev/sound/pcm/sound.c pcm_setstatus
ANALYZE: /home/s/projects/dailybuild/src/sys/dev/sound/pcm/sound.c pcm_getflag3 diagnostics generated.
s
ANALYZE: /home/s/projects/dailybuild/src/sys/dev/sound/pcm/sound.c pcm_setflags
ANALYZE: /home/s/projects/dailybuild/src/sys/dev/sound/pcm/sound.c pcm_getdevinfo

It doesn't happen too often. Of 2712 "x diagnostics generated" messages in my log, 321 are not at the beginning of the line. I haven't found any pattern, but it always seems to happen for the same files.

If there's any additional info I can provide, please tell me.

Regards,
Sascha

#2

Sascha Wildner schrieb:

Hi,

I've noticed that occasionally the "x diagnostics generated" output is mixed into "ANALYZE: ..." lines in scan-build output, as in the following snippet:

Sorry, the formatting was probably not optimal, so here it is again (hopefully better):

ANALYZE: /home/s/projects/dailybuild/src/sys/dev/sound/pcm/sound.c pcm_chn_remove
ANALYZE: /home/s/projects/dailybuild/src/sys/dev/sound/pcm/sound.c pcm_addchan
ANALYZE: /home/s/projects/dailybuild/src/sys/dev/sound/pcm/sound.c pcm_killchan
ANALYZE: /home/s/projects/dailybuild/src/sys/dev/sound/pcm/sound.c pcm_setstatus
ANALYZE: /home/s/projects/dailybuild/src/sys/dev/sound/pcm/sound.c pcm_getflag3 diagnostics generated.
s
ANALYZE: /home/s/projects/dailybuild/src/sys/dev/sound/pcm/sound.c pcm_setflags
ANALYZE: /home/s/projects/dailybuild/src/sys/dev/sound/pcm/sound.c pcm_getdevinfo

#3

This might be due to some recent changes to the diagnostics machinery and the flushing of stream buffers. I'll try and look into this soon.

#4

Hi Sascha,

I'm hoping that the following patch will address the issue:

http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20081117/009553.html

Please let me know if you continue to see such output.

Ted

#5

Ted Kremenek schrieb:

Hi Sascha,

I'm hoping that the following patch will address the issue:

http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20081117/009553.html

Please let me know if you continue to see such output.

Ted,

all messages are at the right place now, thanks.

Another issue I'm seeing are occasional messages a la

Error opening output file '.../clang_crash_2TI9h0.i'

even though the files in question seem to be created just fine.

Regards,
Sascha