[LTP] [PATCH v3] Add process_madvise01 test

[Petr Vorel]

> Hi!
> > > > +	TST_EXP_EXPR(map_before.swap < map_after.swap,
> > > > +		"Most of the memory has been swapped out: %dkB out of %dkB",
> > > > +		map_after.swap,

> > > Although it's unlikely that map_before.swap will be non-zero we should
> > > print the difference here, so I've changed this part to:

> > > map_after.swap - map_before.swap

> > > And pushed, thanks.

> > FYI this new test is failing on current openSUSE Tumbleweed (kernel 6.4.12-1):

> > tst_test.c:1559: TINFO: Timeout per run is 0h 00m 30s
> > process_madvise01.c:38: TINFO: Allocate memory: 1048576 bytes
> > process_madvise01.c:99: TINFO: Reclaim memory using MADV_PAGEOUT
> > process_madvise01.c:62: TFAIL: Expect: Most of the memory has been swapped out: 0kB out of 1024kB

> > Summary:
> > passed   0
> > failed   1
> > broken   0
> > skipped  0
> > warnings 0

> > Test works well on current Debian unstable with kernel 6.4.11.
> > Could you please have a look if it's a test or a kernel bug?

> Does the machine where it fails even have any swap?

I has CONFIG_SWAP=y, but vmstat reports si and so as 0.
Therefore machine probably does not have any swap.

> I suppose that we should check for:

> 1) CONFIG_SWAP in the kernel .config
> 2) /proc/meminfo SwapFree: > $pagesize kB

+1.

> I guess that the madvise kicks off the kernel swapper, but it may take
> some time for the memory to be actually swapped.

Sounds reasonable.

BTW I wonder if we should add 'vmstat' or 'cat /proc/meminfo' to ver_linux.

Kind regards,
Petr