kern/52409: uvm_fault when waking up after sleep (Let's Note CF-{N,S}-9, etc.)

[makoto%ki.nu@localhost]

>Number:         52409
>Category:       kern
>Synopsis:       uvm_fault when waking up after sleep (Let's Note CF-{N,S}-9, etc.)
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Jul 16 04:50:00 +0000 2017
>Originator:     Makoto Fujiwara
>Release:        NetBSD 8.99.1
>Organization:
KINU Corporation

>Environment:
	
	
NetBSD 8.99.1 (GENERIC) #0: Sat Jul 15 10:17:09 UTC 2017
        mkrepro%mkrepro.NetBSD.org@localhost:/usr/src/sys/arch/amd64/compile/GENERIC

Architecture: x86_64
Machine: amd64

>Description:
At waking up after sleep state, 

(1) In file sys/external/bsd/drm2/pci/drm_pci.c, 
   ---------------------
   234  static int
   235  drm_pci_irq_install(struct drm_device *dev, irqreturn_t (*handler)(void *),
   236      int flags, const char *name, void *arg, struct drm_bus_irq_cookie **cookiep)

        ... (omit) ...
   253          intrstr = pci_intr_string(pa->pa_pc, irq_cookie->intr_handles[0],
   254              intrbuf, sizeof(intrbuf));
   ---------------------
   at the line 253, irq_cookie->intr_handles == 0, 
   which causes following trap.

    acpi0: entering state S3
    Flushing disk caches: 10 done
    ioapic0 reenabling
    uvm_fault(0xffffe40110dc68b8, 0x0, 1) -> e
    fatal page fault in supervisor mode
    trap type 6 code 0 rip 0xffffffff806435c8 cs 0x8 rflags 0x10202 cr2 0 ilevel 0 rsp 0xffffe4004003baa0
    curlwp 0xffffe40115c38480 pid 823.1 lowest kstack 0xffffe400400382c0
    panic: trap
    cpu0: Begin traceback...
    vpanic() at netbsd:vpanic+0x140
    snprintf() at netbsd:snprintf
    trap() at netbsd:trap+0xc6b
    --- trap (number 6) ---
    drm_pci_irq_install() at netbsd:drm_pci_irq_install+0x90
    drm_irq_install() at netbsd:drm_irq_install+0xb7
    __i915_drm_thaw() at netbsd:__i915_drm_thaw+0xc9
    i915drmkms_resume() at netbsd:i915drmkms_resume+0x34
    device_pmf_driver_resume() at netbsd:device_pmf_driver_resume+0x46
    pmf_device_resume_locked() at netbsd:pmf_device_resume_locked+0xd3
    pmf_device_resume() at netbsd:pmf_device_resume+0x41
    pmf_system_resume() at netbsd:pmf_system_resume+0xc7
    acpi_enter_sleep_state() at netbsd:acpi_enter_sleep_state+0x182
    sysctl_hw_acpi_sleepstate() at netbsd:sysctl_hw_acpi_sleepstate+0xfe
    sysctl_dispatch() at netbsd:sysctl_dispatch+0xc1
    sys___sysctl() at netbsd:sys___sysctl+0xd8
    syscall() at netbsd:syscall+0x1d8
    --- syscall (number 202) ---
    727e27b1bb6a:
    cpu0: End traceback...
    
    dumping to dev 0,1 (offset=8802119, size=997005):
    dump Copyright (c) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005,
        2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017
        The NetBSD Foundation, Inc.  All rights reserved.
    Copyright (c) 1982, 1986, 1989, 1991, 1993
        The Regents of the University of California.  All rights reserved.
    
    NetBSD 8.99.1 (GENERIC) #0: Sat Jul 15 10:17:09 UTC 2017
    	mkrepro%mkrepro.NetBSD.org@localhost:/usr/src/sys/arch/amd64/compile/GENERIC
    total memory = 3894 MB
    avail memory = 3759 MB

>How-To-Repeat:
	
1. Have the machine, Let's Note CF-N9 or CF-S9. CF-J10.
   (SX2 or N10 are OK for this purpose, but they have another
   problem)

2. Install 8.99.1 from the daily 
   http://nycdn.netbsd.org/pub/NetBSD-daily/HEAD/201707151110Z/amd64/

3. Replace lid_switch by sleep_button in followin path.

   ls -lt /etc/powerd/scripts/{sl*,lid*}
   -r-xr-xr-x  1 root  wheel  1035 Jul  8 00:41 /etc/powerd/scripts/lid_switch
   -r-xr-xr-x  1 root  wheel   513 Mar 12 03:45 /etc/powerd/scripts/lid_switch.orig
   -r-xr-xr-x  1 root  wheel  1035 Mar 12 03:45 /etc/powerd/scripts/sleep_button
 
4. Boot the machine

5. When you see login: prompt, close the lid
   (PC gets sleep) .. this works only once after reboot.
   see another PR
   42051 lid switch works only once
   49024 Powerd invokes lid_switch by acpilid only ONCE after boot

6. Wait few seconds, and open the lid

7. You may see some lines for trap or panic (sometimes may not), and soon
   screen gets dark out and one or two mins later, it will reboot automatically

Note: (a)  in above, don't set ddb.onpanic to 1.
Note: (b)  You may replace the procedure 3 to 5 above, 
           login and do
             sudo sysctl -w hw.acpi.sleep.state=3
           may give the equivalent effect.

>Fix:
	

diff --git a/sys/external/bsd/drm2/i915drm/i915_pci.c b/sys/external/bsd/drm2/i915drm/i915_pci.c
index 3ba85bc531c..5b46e472102 100644
--- a/sys/external/bsd/drm2/i915drm/i915_pci.c
+++ b/sys/external/bsd/drm2/i915drm/i915_pci.c
@@ -241,6 +241,7 @@ i915drmkms_resume(device_t self, const pmf_qual_t *qual)
 
 	if (dev == NULL)
 		return true;
+	pci_enable_msi(dev->pdev);
 
 	ret = i915_drm_thaw_early(dev);
 	if (ret)

>Unformatted: