Testing in the Yocto Project

Posted May 20, 2019 9:16 UTC (Mon) by metan (subscriber, #74107)
In reply to: Testing in the Yocto Project by roc
Parent article: Testing in the Yocto Project

> Indeed, the rr test suite still catches kernel regressions from time to time, which we investigate and report. I don't think anyone else is running systematic tests of x86-32 ioctl compat code.

Well in SUSE we do run LTP compiled with -m32 on x86-64 which catches bugs in the compat layer from time to time, but the coverage of ioctls() is sparse at best, we do have some tests for /dev/tty*, /dev/net/tun, block device ioctls, /dev/random, namespace ioctls and some Btrfs ones and that's it.

Testing in the Yocto Project

Posted May 20, 2019 13:14 UTC (Mon) by roc (subscriber, #30627) [Link] (1 responses)

One thing we're doing which LTP probably isn't doing: we make syscall parameter buffers be immediately followed by an unmapped guard page, so if the kernel reads or writes too much memory, the test gets an EFAULT. Bugs where the kernel reads too much memory (and ignores what was read) are otherwise impossible to catch. We do this on x86-64 too of course but it catches fewer bugs there.