ldv changed the topic of #strace to: https://strace.io | https://strace.io/logs/ | strace 4.23 is out | strace-devel@lists.strace.io for dev discussions
pombreda has joined #strace
mjw has joined #strace
<zhibli> Hi, I'm trying to write test for ioctl_evdev but I'm stuck now.
<zhibli> Because it seems ioctl(fd, EVIOCGBIT(EV_SYN, EV_MAX), arg) cannot be decoded correctly.
<zhibli> Should EVIOCGBIT(0, n) be decoded to get event types or the corresponding set of codes?
<zhibli> I'm not so familiar with this and I'm still learning this confusing part.
<zhibli> As I see, strace tries to get the corresponding set of codes of the specific type such as EV_LED, EV_KEY.
<zhibli> All those can be decoded correctly except EV_SYN.
<esyrd> yes, there's no EV_SYN in EVIOCGBIT, it's a bug in decoder.
<zhibli> I tried to find issues of this on github but I didn't see one
<esyrd> what issues?
<esyrd> and why there should be anything?
<zhibli> Oh I thought the all the known bugs would be recorded there
<esyrd> well, this one wasn't known previously, and github is just a way for some to report something, not that we prefer to use it (as it avoids discussion being recorded into the mailing list archive).
<zhibli> OK I see
<esyrd> as a matter of fact, there's no single repository of bugs, unfortunately.
<zhibli> well maintaining a extra bug tracking system like bugzilla requires too much right?
<esyrd> so, yep, you're encouraged to fix issues and mistakes with evdev decoder, it definitely needs some love.
<esyrd> not that too much, but it still requires effort, and we don't have surplus of it.
<zhibli> Got it, I may have to do more research about this to see how to fix it.
<zhibli> Since event ioctl is something new to me
<esyrd> Well, drivers/input/evdev.c is not too big, I would say.
<zhibli> OK I will try. So I can temporarily skip this particular part and continue to write tests for other part.
shaunP has joined #strace
<shaunP> Hi. A couple years back, Linux added "kcov" which allows a user to extract per task code coverage. 6 or so months ago, I extended strace to capture per-system call code coverage along with system call information part of a research project. Is this a feature that seems interesting to you all. The current implementation is pretty messy but can be found here https://github.com/shankarapailoor/strace.
<shaunP> If this is a feature that seems interesting to you all, I would love to work to get it up to your standards. I may have shown this to a few of you a while back when I encountered some issues
shaunP has quit [Ping timeout: 252 seconds]
shaunP has joined #strace
jp has quit [Ping timeout: 256 seconds]
jp has joined #strace
shaunP has quit [Ping timeout: 252 seconds]
mjw has quit [Quit: zzz]