[PATCH v8 03/11] app/test: replace POSIX specific code

[Jie Zhou]

On Wed, Nov 24, 2021 at 01:02:06AM +0300, Dmitry Kozlyuk wrote:
> 2021-10-25 19:45 (UTC-0700), Jie Zhou:
> > - Include rte_os_shim.h
> > - Replace sleep and usleep with rte_delay_us_sleep
> > - #ifndef RTE_EXEC_ENV_WINDOWS for POSIX code only
> > 
> > Signed-off-by: Jie Zhou <jizh at linux.microsoft.com>
> > ---
> 
> This patch can be combined with the previous one:
> they serve the same purpose---to remove Unix-specific code.
> 
> Please try to summarize in the commit message
> which parts of the tests suites are excluded, e.g. multi-process.
> It is more useful then stating what was changed in the code.
>
Will combine and revise the message.
 
> [...]
> > diff --git a/app/test/test_cmdline.c b/app/test/test_cmdline.c
> > index 115bee966d..9a76bd299f 100644
> > --- a/app/test/test_cmdline.c
> > +++ b/app/test/test_cmdline.c
> > @@ -31,6 +31,7 @@ test_cmdline(void)
> >  		return -1;
> >  	if (test_parse_num_invalid_param() < 0)
> >  		return -1;
> > +#ifndef RTE_EXEC_ENV_WINDOWS
> >  	printf("Testing parsing IP addresses...\n");
> >  	if (test_parse_ipaddr_valid() < 0)
> >  		return -1;
> > @@ -38,6 +39,7 @@ test_cmdline(void)
> >  		return -1;
> >  	if (test_parse_ipaddr_invalid_param() < 0)
> >  		return -1;
> > +#endif
> >  	printf("Testing parsing strings...\n");
> >  	if (test_parse_string_valid() < 0)
> >  		return -1;
> 
> What's wrong with parsing IP addresses on Windows?
> 
test_cmdline_ipaddr.c uses linux netinet/in.h specific u6_addr. Skip these 3 cases for now and prefer a separate patch to make it work on Windows. Or maybe there is already DPDK support on this which I am not aware of? Thanks.

> [...]
> > diff --git a/app/test/test_mp_secondary.c b/app/test/test_mp_secondary.c
> > index 5b6f05dbb1..da035348bd 100644
> > --- a/app/test/test_mp_secondary.c
> > +++ b/app/test/test_mp_secondary.c
> > @@ -14,7 +14,9 @@
> >  #include <errno.h>
> >  #include <string.h>
> >  #include <unistd.h>
> > +#ifndef RTE_EXEC_ENV_WINDOWS
> >  #include <sys/wait.h>
> > +#endif
> >  #include <libgen.h>
> >  #include <dirent.h>
> >  #include <limits.h>
> 
> <libgen.h> is absent on Windows for sure, but you don't exclude it.
> Does this file even need modification?
> It's not going to be compiled for Windows.
> 
This is replaced with the test stub in the patch#11 (of V8) to make it compile on Windows. Sorry for the confusion. Will make sure remove this unnecessary part from V9.

> [...]
> > diff --git a/app/test/test_ring_stress.c b/app/test/test_ring_stress.c
> > index 1af45e0fc8..ce3535c6b2 100644
> > --- a/app/test/test_ring_stress.c
> > +++ b/app/test/test_ring_stress.c
> > @@ -43,9 +43,10 @@ test_ring_stress(void)
> >  	n += test_ring_rts_stress.nb_case;
> >  	k += run_test(&test_ring_rts_stress);
> >  
> > +#ifndef RTE_EXEC_ENV_WINDOWS
> >  	n += test_ring_hts_stress.nb_case;
> >  	k += run_test(&test_ring_hts_stress);
> > -
> > +#endif
> 
> Can you please elaborate what is the issue with this case?
> It is also one of the details you usually want to put
> into the commit message.

Cannot remember what caused this case being skipped in the first place, but it can work now. So removed the ifndef. Thanks.