ddcutil core dump when DSA2 cache file not found/created about ddcutil HOT 4 CLOSED

@agspoon Thank you for the detailed report. Will follow up.

Why are you running ddcutil from a udev rule? What's the use case? I continue to be surprised by the ways people use ddcutil. Other than ddcutil chkusbmon which is specially coded to be run from a udev rule, ddcutil assumes a normally configured environment.

Options --disable-capabilities-cache and --disable-dynamic-sleep should eliminate file io, and --syslog never will disable writing to the system log.

As for suppressing an error message if a file cannot be written, I believe that by default things should not fail silently. I think its reasonable to expect require use of command line options to disable the features that cause the error messages.

from ddcutil.

ddcutil now checks for NULL values returned by dsa2_stats_cache_file_name() and related functions. which should resolve the segfault problem. The changes have been pushed to branch 2.0.0-dev.

If NULL values are encountered, ddcutil writes error messages to the terminal and system log. Use --disable-dynamic-sleep etc. to avoid them.

from ddcutil.

Thanks! User's, the bane of every developer :)

My use case is basically that I'm too cheap to buy a KVM to switch between my two computers. A low level aggravation that I suffer a couple dozen times a day. I have a small USB A/B switch that I use to swap the keyboard/mouse, but had to go through a awkward button and OSD interface on my LG monitor to switch inputs.

Thanks to someone figuring out how to programmatically switch inputs on these broken LG monitors (Issue #100) , and your implementation of the --i2c-source-addr option, I can now automate the process.

I wrote a couple udev rules that fire whenever the USB keyboard is added or removed, that runs ddcutil to switch the monitor input accordingly. Works great. A poor man's KVM. I noticed only later that my journal was filled with backtrace logs from the above SEGV fault in ddcutil (apparently the stats are written after the work is done).

I agree that suppressing the error message is not a good idea, I was just wondering if I was missing out on some important functionality by disabling the dynamic-sleep capability, when all I really wanted to avoid was writing to the cache file. Regardless, it works for my weird corner use case, so I'm good. Just thought I'd pass on what I found regarding the NULL pointer.

Thanks again :)

from ddcutil.

Thank you for letting me know the use case. I'll keep it in mind as I'm working on ddcutil. I'm continually amazed at the clever ways people find to use ddcutil.

Dynamic sleep depends on having a cache file to tune the sleep-multiplier over many program executions, so disabling dynamic sleep has little impact on execution time in your case. If execution time matters, you can experiment with option --sleep-multiplier.

from ddcutil.