Re: [PATCH 00/12] linux-user: Add support for real time clock and

[Laurent Vivier <laurent@vivier.eu>]

Le 14/01/2020 à 13:39, Philippe Mathieu-Daudé a écrit :
> Hi Filip,
> 
> This is v5.
> 
> On 1/9/20 1:59 PM, Filip Bozuta wrote:
>> MIME-Version: 1.0
>> Content-Type: text/plain; charset=UTF-8
>> Content-Transfer-Encoding: 8bit
>>
>> This series covers following RTC and sound timer ioctls:
>>
>>    RTC ioctls(22):
>>
>>      * RTC_AIE_ON          * RTC_ALM_SET         * RTC_WKALM_SET
>>      * RTC_AIE_OFF         * RTC_ALM_READ        * RTC_WKALM_RD
>>      * RTC_UIE_ON          * RTC_RD_TIME         * RTC_PLL_GET
>>      * RTC_UIE_OFF         * RTC_SET_TIME        * RTC_PLL_SET
>>      * RTC_PIE_ON          * RTC_IRQP_READ       * RTC_VL_READ
>>      * RTC_PIE_OFF         * RTC_IRQP_SET        * RTC_VL_CLR
>>      * RTC_WIE_ON          * RTC_EPOCH_READ
>>      * RTC_WIE_OFF         * RTC_EPOCH_SET
>>
>>    Sound timer ioctls(14):
>>
>>      * SNDRV_TIMER_IOCTL_PVERSION          * SNDRV_TIMER_IOCTL_INFO
>>      * SNDRV_TIMER_IOCTL_NEXT_DEVICE       * SNDRV_TIMER_IOCTL_PARAMS
>>      * SNDRV_TIMER_IOCTL_TREAD             * SNDRV_TIMER_IOCTL_STATUS
>>      * SNDRV_TIMER_IOCTL_GINFO             * SNDRV_TIMER_IOCTL_START
>>      * SNDRV_TIMER_IOCTL_GPARAMS           * SNDRV_TIMER_IOCTL_STOP
>>      * SNDRV_TIMER_IOCTL_GSTATUS           * SNDRV_TIMER_IOCTL_CONTINUE
>>      * SNDRV_TIMER_IOCTL_SELECT            * SNDRV_TIMER_IOCTL_PAUSE
>>
>> The functionalities of individual ioctls were described in this series
>> patch commit messages.
>>
>> Testing method for RTC ioctls:
>>
>>      Mini test programs were written for each ioctl. Those programs were
>>      compiled (sometimes using cross-compilers) for the following
>>      architectures:
>>
>>          * Intel 64-bit (little endian)
>>          * Power pc 32-bit (big endian)
>>          * Power pc 64-bit (big endian)
>>
>>      The corresponding native programs were executed without using
>>      QEMU on following hosts:
>>
>>          * Intel Core i7-4790K (x86_64 host)
>>          * Power 7447A (ppc32 host)
>>
>>      All applicable compiled programs were in turn executed through QEMU
>>      and the results obtained were the same ones gotten for native
>>      execution.
>>
>>      Example of a test program:
>>
>>          For ioctl RTC_RD_TIME the following test program was used:
>>
>>          #include <stdio.h>
>>          #include <stdlib.h>
>>          #include <linux/rtc.h>
>>          #include <fcntl.h>
>>          #include <linux/input.h>
>>          #include <sys/types.h>
>>          #include <unistd.h>
>>
>>          #define ERROR -1
>>
>>          int main()
>>          {
>>
>>              int fd = open("/dev/rtc", O_RDWR | O_NONBLOCK);
>>
>>              if(fd == ERROR)
>>              {
>>                  perror("open");
>>                  return -1;
>>              }
>>
>>              struct rtc_time cur_time;
>>
>>              if(ioctl(fd, RTC_RD_TIME, &cur_time) < 0)
>>              {
>>                  perror("ioctl");
>>                  return -1;
>>              }
>>
>>              printf("Second: %d, Minute: %d, Hour: %d, Day: %d, Month:
>> %d, Year: %d,",
>>                  cur_time.tm_sec, cur_time.tm_min, cur_time.tm_hour,
>>                  cur_time.tm_mday, cur_time.tm_mon, cur_time.tm_year);
>>
>>              return 0;
>>          }
>>
>>      Limitations of testing:
>>
>>          The test host pc that was used for testing (intel pc) has RTC
>>          that doesn't support all RTC features that are accessible
>>          through ioctls. This means that testing can't discover
>>          functionality errors related to the third argument of ioctls
>>          that are used for features which are not supported. For example,
>>          running the test program for ioctl RTC_EPOCH_READ gives
>>          the error output: inappropriate ioctl for device. As expected,
>>          the same output was obtained through QEMU which means that this
>>          ioctl is recognized in QEMU but doesn't really do anything
>>          because it is not supported in the host computer's RTC.
>>
>>          Conclusion: Some RTC ioctls need to be tested on computers
>>          that support their functionalities so that it can be inferred
>>          that they are really supported in QEMU. In absence of such
>>          test hosts, the specifications of those ioctls need to be
>>          carefully checked manually and the implementations should be
>>          updated accordingly.
> 
> We should be able to check if the host supports these features and run
> your different tests.
> Can you add them in the repository? Maybe under tests/tcg/multiarch/.
> 

I agree with Philippe. If you provide the test we will also be able to
run non-regression tests. You can add them in a separate patch if it's
easier for you.

Thanks,
Laurent