Note this could be duplicate of BUG 102490.

AFAICS, in the tests the 500Hz interrupt frequency is too high for the main thread, so syscalls there will be interrupted and delayed indefinitely leading to watchdog timeout. To test this theory we should try something like:

diff --git a/lib/igt_aux.c b/lib/igt_aux.c
index fa6594c3..2f01c723 100644
--- a/lib/igt_aux.c
+++ b/lib/igt_aux.c
@@ -278,7 +278,7 @@ static void __attribute__((noreturn)) signal_helper_process(pid_t pid)
 {
        /* Interrupt the parent process at 500Hz, just to be annoying */
        while (1) {
-               usleep(1000 * 1000 / 500);
+               usleep(1000 * 1000 / 50);
                if (kill(pid, SIGCONT)) /* Parent has died, so must we. */
                        exit(0);
        }

The softdog fires when the kernel stops making progress. Constantly getting back to userspace is progress from the kernel's pov.

It would result in the test not completing, but that's not what's going on here.

This is what I get:

ickle@broxton:~/intel-gpu-tools/tests$ sudo ./kms_flip --r vblank-vs-dpms-suspend-interruptible
IGT-Version: 1.20-gd8493ea7 (x86_64) (Linux: 4.14.0-rc4+ x86_64)
Using monotonic timestamps
Beginning vblank-vs-dpms-suspend-interruptible on pipe A, connector DP-2
  1920x1080 60 1920 2008 2052 2200 1080 1084 1089 1125 0x5 0x48 148500
.rtcwake: wakeup from "mem" using /dev/rtc0 at Wed Oct 11 18:58:44 2017

Message from syslogd@broxton at Oct 11 19:58:44 ...
 kernel:[24901.110247] mce: [Hardware Error]: CPU 0: Machine Check: 0 Bank 4: a600000000020408

Message from syslogd@broxton at Oct 11 19:58:44 ...
 kernel:[24901.110251] mce: [Hardware Error]: TSC 0 ADDR fef13b00 

Message from syslogd@broxton at Oct 11 19:58:44 ...
 kernel:[24901.110267] mce: [Hardware Error]: PROCESSOR 0:506c9 TIME 1507748324 SOCKET 0 APIC 0 microcode 20
(kms_flip:31975) CRITICAL: Test assertion failure function run_test_step, file kms_flip.c:750:
(kms_flip:31975) CRITICAL: Failed assertion: (drmWaitVBlank(drm_fd, &vbl)) == 0
(kms_flip:31975) CRITICAL: Last errno: 22, Invalid argument
Stack trace:
  #0 [__igt_fail_assert+0xf1]
  #1 [run_test_on_crtc_set.constprop.14+0x1d36]
  #2 [run_test+0x272]
  #3 [main+0x492]
  #4 [__libc_start_main+0xf1]
  #5 [_start+0x2a]
  #6 [<unknown>+0x2a]
Subtest vblank-vs-dpms-suspend-interruptible failed.
**** DEBUG ****
(kms_flip:31975) DEBUG: Test requirement passed: (flags & TEST_HANG) == 0 || !is_wedged(drm_fd)
(kms_flip:31975) DEBUG: Test requirement passed: modes
(kms_flip:31975) INFO: Beginning vblank-vs-dpms-suspend-interruptible on pipe A, connector DP-2
(kms_flip:31975) igt-fb-DEBUG: igt_create_fb_with_bo_size(width=1920, height=1080, format=0x34325258, tiling=0x0, size=0)
(kms_flip:31975) igt-fb-DEBUG: igt_create_fb_with_bo_size(handle=2, pitch=7680)
(kms_flip:31975) igt-fb-DEBUG: igt_create_fb_with_bo_size(width=1920, height=1080, format=0x34325258, tiling=0x0, size=0)
(kms_flip:31975) igt-fb-DEBUG: igt_create_fb_with_bo_size(handle=3, pitch=7680)
(kms_flip:31975) igt-kms-INFO:   1920x1080 60 1920 2008 2052 2200 1080 1084 1089 1125 0x5 0x48 148500
(kms_flip:31975) igt-core-DEBUG: Test requirement passed: !igt_run_in_simulation()
(kms_flip:31975) igt-aux-DEBUG: Test requirement passed: (power_dir = open("/sys/power", O_RDONLY)) >= 0
(kms_flip:31975) igt-aux-DEBUG: Test requirement passed: get_supported_suspend_states(power_dir) & (1 << state)
(kms_flip:31975) igt-aux-DEBUG: Test requirement passed: test == SUSPEND_TEST_NONE || faccessat(power_dir, "pm_test", R_OK | W_OK, 0) == 0
(kms_flip:31975) igt-aux-DEBUG: Test requirement passed: ret == 0
(kms_flip:31975) DEBUG: name = vblank
last_ts = 0.000000
last_received_ts = 0.000000
last_seq = 0
current_ts = 24901.122355
current_received_ts = 24901.976559
current_seq = 11
count = 0
seq_step = 10
(kms_flip:31975) CRITICAL: Test assertion failure function run_test_step, file kms_flip.c:750:
(kms_flip:31975) CRITICAL: Failed assertion: (drmWaitVBlank(drm_fd, &vbl)) == 0
(kms_flip:31975) CRITICAL: Last errno: 22, Invalid argument
(kms_flip:31975) igt-core-INFO: Stack trace:
(kms_flip:31975) igt-core-INFO:   #0 [__igt_fail_assert+0xf1]
(kms_flip:31975) igt-core-INFO:   #1 [run_test_on_crtc_set.constprop.14+0x1d36]
(kms_flip:31975) igt-core-INFO:   #2 [run_test+0x272]
(kms_flip:31975) igt-core-INFO:   #3 [main+0x492]
(kms_flip:31975) igt-core-INFO:   #4 [__libc_start_main+0xf1]
(kms_flip:31975) igt-core-INFO:   #5 [_start+0x2a]
(kms_flip:31975) igt-core-INFO:   #6 [<unknown>+0x2a]
****  END  ****
Subtest vblank-vs-dpms-suspend-interruptible: FAIL (0.960s)

Hmm, chvt broke after running that test:

[<ffffffff813513ac>] __vt_event_wait.isra.0.part.1+0x2c/0x60
[<ffffffff813515ac>] vt_waitactive+0x6c/0xc0
[<ffffffff813522c8>] vt_ioctl+0xb78/0x1220
[<ffffffff8134592b>] tty_ioctl+0xdb/0x850
[<ffffffff81193ddf>] do_vfs_ioctl+0x8f/0x5b0
[<ffffffff8119433c>] SyS_ioctl+0x3c/0x70
[<ffffffff81538660>] entry_SYSCALL_64_fastpath+0x13/0x94

Here is another APL-shards softdog

https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3210/shard-apl6/igt@gem_cpu_reloc@full.html

Commit in IGT:
author	Imre Deak <imre.deak@intel.com>	2017-10-13 11:30:09 (GMT)
committer	Imre Deak <imre.deak@intel.com>	2017-10-16 10:35:41 (GMT)
commit	2aeececacc18204f763ec946b35877afb1f8f5b5

Imre patch was integrated in CI_DRM_3241 since then the incompletes on APL-shards on igt@kms_flip@vblank-vs-dpms-suspend-interruptible,
igt@kms_flip@vblank-vs-modeset-suspend-interruptible and
igt@kms_flip@vblank-vs-suspend-interruptible has not been reproduced.