[PATCH v4] testpmd: cleanup cleanly from signal

[Mattias Rönnblom]

On 2022-11-09 23:53, Stephen Hemminger wrote:
> On Wed, 9 Nov 2022 22:46:55 +0100
> Mattias Rönnblom <hofors at lysator.liu.se> wrote:
> 
>> On 2022-11-09 05:10, Stephen Hemminger wrote:
>>> Do a clean shutdown of testpmd when a signal is received;
>>> instead of having testpmd kill itself.
>>> This fixes problem where a signal could be received
>>> in the middle of a PMD and then the signal handler would call
>>> PMD's close routine which could cause a deadlock.
>>>
>>> Added benefit is it gets rid of Windows specific code.
>>>
>>> Fixes: d9a191a00e81 ("app/testpmd: fix quitting in container")
>>> Signed-off-by: Stephen Hemminger <stephen at networkplumber.org>
>>> ---
>>> v4 - use select() because that is available on Windows; and other
>>>        functions poll() and sigaction() are not.
>>>
>>>    app/test-pmd/testpmd.c | 63 +++++++++++++++++++++++-------------------
>>>    1 file changed, 34 insertions(+), 29 deletions(-)
>>>
>>> diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
>>> index cf5942d0c422..274e96cac2d4 100644
>>> --- a/app/test-pmd/testpmd.c
>>> +++ b/app/test-pmd/testpmd.c
>>> @@ -12,6 +12,7 @@
>>>    #ifndef RTE_EXEC_ENV_WINDOWS
>>>    #include <sys/mman.h>
>>>    #endif
>>> +#include <sys/select.h>
>>>    #include <sys/types.h>
>>>    #include <errno.h>
>>>    #include <stdbool.h>
>>> @@ -4251,26 +4252,11 @@ print_stats(void)
>>>    static void
>>>    signal_handler(int signum)
>>>    {
>>> -	if (signum == SIGINT || signum == SIGTERM) {
>>> -		fprintf(stderr, "\nSignal %d received, preparing to exit...\n",
>>> -			signum);
>>> -#ifdef RTE_LIB_PDUMP
>>> -		/* uninitialize packet capture framework */
>>> -		rte_pdump_uninit();
>>> -#endif
>>> -#ifdef RTE_LIB_LATENCYSTATS
>>> -		if (latencystats_enabled != 0)
>>> -			rte_latencystats_uninit();
>>> -#endif
>>> -		force_quit();
>>> -		/* Set flag to indicate the force termination. */
>>> -		f_quit = 1;
>>> -		/* exit with the expected status */
>>> -#ifndef RTE_EXEC_ENV_WINDOWS
>>> -		signal(signum, SIG_DFL);
>>> -		kill(getpid(), signum);
>>> -#endif
>>> -	}
>>> +	fprintf(stderr, "\nSignal %d %s received, preparing to exit...\n",
>>> +		signum, strsignal(signum));
>>
>> fprintf() is not async signal safe, and neither is strsignal().
>>
>> This is not a regression introduced by this patch, but I thought it
>> might be worth fixing.
>>
>>> +
>>> +	/* Set flag to indicate the force termination. */
>>> +	f_quit = 1;
>>>    }
>>>    
>>>    int
>>> @@ -4449,9 +4435,6 @@ main(int argc, char** argv)
>>>    	} else
>>>    #endif
>>>    	{
>>> -		char c;
>>> -		int rc;
>>> -
>>>    		f_quit = 0;
>>>    
>>>    		printf("No commandline core given, start packet forwarding\n");
>>> @@ -4476,15 +4459,37 @@ main(int argc, char** argv)
>>>    				prev_time = cur_time;
>>>    				rte_delay_us_sleep(US_PER_S);
>>>    			}
>>> -		}
>>> +		} else {
>>> +			char c;
>>> +			fd_set fds;
>>>    
>>> -		printf("Press enter to exit\n");
>>> -		rc = read(0, &c, 1);
>>> -		pmd_test_exit();
>>> -		if (rc < 0)
>>> -			return 1;
>>> +			printf("Press enter to exit\n");
>>> +
>>> +			FD_ZERO(&fds);
>>> +			FD_SET(0, &fds);
>>> +
>>> +			if (select(1, &fds, NULL, NULL, NULL) <= 0) {
>>> +				fprintf(stderr, "Select failed: %s\n",
>>> +					strerror(errno));
>>
>> Why is select() needed? Wouldn't a blocking read suffice? Or getchar().
> 
> On Linux, signal set SA_RESTART so a simple read is not interrupted.
> One option was to use sigaction() which allows controlling flags, but that
> won't work on Windows.  Using select() works on both.
> 

OK, so select() is used because a signal might interrupt read() on Windows?

while (read(0, &c, 1) == -1 && errno == EINTR)
         ;

Would that work?

(select() won't return 0 since you don't have a timeout.)