More splice() testcases

[metan-ucw]

Recently there were some change in splice() kernel implementation that broke a few cases we did not even consider to test.

commit 14e3e989f6a5d9646b6cf60690499cc8bdc11f7d Author: Linus Torvalds [email protected] Date: Sun Dec 27 10:56:33 2020 -0800

proc mountinfo: make splice available again

commit 9bb48c82aced07698a2d08ee0f1475a6c4f6b266 Author: Linus Torvalds [email protected] Date: Tue Jan 19 11:41:16 2021 -0800

tty: implement write_iter

We should add coverage at least for splice() on proc files and possibly the tty as well.

[richiejp]

Perhaps we need a "write any FD to and FD", test. Which tests copying data between objects addressable by File Descriptors using various methods.

Some other System calls to consider: readv/writev, tee, vmsplice, mmap

[coolgw]

First case for splice on proc https://patchwork.ozlabs.org/project/ltp/patch/[email protected]/

[coolgw]

https://patchwork.ozlabs.org/project/ltp/patch/[email protected]/ accept.