Re: kern/50060: kernel crash with i915drmksm on Intel 965Q

[Andreas Gustafsson <gson%gson.org@localhost>]

The following reply was made to PR kern/50060; it has been noted by GNATS.

From: Andreas Gustafsson <gson%gson.org@localhost>
To: gnats-bugs%NetBSD.org@localhost, christos%NetBSD.org@localhost
Cc: 
Subject: Re: kern/50060: kernel crash with i915drmksm on Intel 965Q
Date: Mon, 12 Oct 2015 09:51:39 +0300

 Christos wrote:
  There is only one bus_space_unmap() in i915_dma.c can you either put
 > printfs before and after, or comment it out to see if it is there where
 > it fails?
 
 I added the printfs to the 2015.10.10.19.35.15 source, and a couple
 more for context:
 
 Index: i915_dma.c
 ===================================================================
 RCS file: /bracket/repo/src/sys/external/bsd/drm2/dist/drm/i915/i915_dma.c,v
 retrieving revision 1.18
 diff -u -r1.18 i915_dma.c
 --- i915_dma.c	10 Oct 2015 19:35:15 -0000	1.18
 +++ i915_dma.c	11 Oct 2015 18:13:43 -0000
 @@ -133,12 +133,14 @@
  
  	if (ring->status_page.gfx_addr) {
  		ring->status_page.gfx_addr = 0;
 +		printf("pre unmap\n");
  #ifdef __NetBSD__
  		bus_space_unmap(dev->pdev->pd_pa.pa_memt,
  		    dev_priv->dri1.gfx_hws_cpu_bsh, 4096);
  #else
  		iounmap(dev_priv->dri1.gfx_hws_cpu_addr);
  #endif
 +		printf("post unmap\n");		
  	}
  
  	/* Need to rewrite hardware status page */
 @@ -1612,6 +1614,8 @@
  	int ret = 0, mmio_bar, mmio_size;
  	uint32_t aperture_size;
  
 +	printf("i915_driver_load entry\n");
 +
  	info = (struct intel_device_info *) flags;
  
  	/* Refuse to load on gen6+ without kms enabled. */
 @@ -1885,6 +1889,9 @@
  	spin_lock_destroy(&dev_priv->irq_lock);
  #endif
  	kfree(dev_priv);
 +
 +	printf("i915_driver_load exit\n");
 +	
  	return ret;
  }
 
 
 Console output:
 
    agp0 at pchb0: i965-family chipset
    pci_mem_find: expected mem type 00000004, found 00000000
    pci_mem_find: expected mem type 00000004, found 00000000
    agp0: can't find MMIO registers
    ppb0 at pci0 dev 1 function 0: vendor 8086 product 29a1 (rev. 0x02)
    ppb0: PCI Express capability version 1 <Root Port of PCI-E Root Complex> x16 @ 2.5GT/s
    pci1 at ppb0 bus 1
    i915drmkms0 at pci0 dev 2 function 0: vendor 8086 product 29a2 (rev. 0x02)
    i915_driver_load entry
    DRM error in i915_gmch_probe: failed to set up gmch
    uvm_fault(0xffffffff811b27a0, 0x7fc000357000, 2) -> e
    fatal page fault in supervisor mode
    trap type 6 code 2 rip ffffffff80a80c73 cs 8 rflags 10246 cr2 7fc000357000 ilevel 8 rsp ffffffff8137e7c8
    curlwp 0xffffffff8110d740 pid 0.1 lowest kstack 0xffffffff8137b2c0
    kernel: page fault trap, code=0
    Stopped in pid 0.1 (system) at  netbsd:_atomic_swap_64+0x3:     xchgq   %rax,0(%
    rdi)
    db{0}>
 
 Looks like that is not where it fails.
 -- 
 Andreas Gustafsson, gson%gson.org@localhost