Re: kern/48739: Reproducible panic in ld_virtio.c on NetBSD/amd64 guest running under qemu on CentOS 6.5

[Pierre Pronchery <khorben%defora.org@localhost>]

The following reply was made to PR kern/48739; it has been noted by GNATS.

From: Pierre Pronchery <khorben%defora.org@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: kern/48739: Reproducible panic in ld_virtio.c on NetBSD/amd64
 guest running under qemu on CentOS 6.5
Date: Thu, 02 Jul 2015 04:53:19 +0200

 			Hi,
 
 I'm apparently hit by the same thing:
 
 > /netbsd: uvm_fault(0xffffffff810b16c0, 0xffff80008e900000, 1) -> e              
 > /netbsd: fatal page fault in supervisor mode                      
 > /netbsd: trap type 6 code 0 rip ffffffff809cce74 cs 8 rflags 10206 cr2 ffff80008e900ff8 ilevel 8 rsp fffffe810f5097c0
 > /netbsd: curlwp 0xfffffe8216316860 pid 26301.1 lowest kstack 0xfffffe810f5072c0 
 > /netbsd: panic: trap                      
 > /netbsd: cpu1: Begin traceback...                      
 > /netbsd: vpanic() at netbsd:vpanic+0x13c                      
 > /netbsd: snprintf() at netbsd:snprintf                      
 > /netbsd: startlwp() at netbsd:startlwp                      
 > /netbsd: alltraps() at netbsd:alltraps+0x96                      
 > /netbsd: ld_virtio_start() at netbsd:ld_virtio_start+0x14b                      
 > /netbsd: ldstart() at netbsd:ldstart+0x6b                      
 > /netbsd: ldstrategy() at netbsd:ldstrategy+0x101                      
 > /netbsd: bdev_strategy() at netbsd:bdev_strategy+0x5b                           
 > /netbsd: spec_strategy() at netbsd:spec_strategy+0x2c                           
 > /netbsd: VOP_STRATEGY() at netbsd:VOP_STRATEGY+0x33                      
 > /netbsd: genfs_do_io() at netbsd:genfs_do_io+0x195                      
 > /netbsd: genfs_gop_write() at netbsd:genfs_gop_write+0x52                       
 > /netbsd: genfs_do_putpages() at netbsd:genfs_do_putpages+0xbec                  
 > /netbsd: VOP_PUTPAGES() at netbsd:VOP_PUTPAGES+0x3a                      
 > /netbsd: ffs_write() at netbsd:ffs_write+0x354                      
 > /netbsd: VOP_WRITE() at netbsd:VOP_WRITE+0x37                      
 > /netbsd: vn_write() at netbsd:vn_write+0xec                      
 > /netbsd: dofilewrite() at netbsd:dofilewrite+0x97                      
 > /netbsd: sys_write() at netbsd:sys_write+0x5f                      
 > /netbsd: syscall() at netbsd:syscall+0x9a                      
 > /netbsd: --- syscall (number 4) ---   
 
 I am running NetBSD 7.0_RC1 with the tickets "841, 847, 849, 853, 855"
 applied.
 
 It seems to happen right after this call:
 
 > 363         bus_dmamap_sync(vsc->sc_dmat, vr->vr_payload,
 > 364                         0, bp->b_bcount,
 > 365                         isread?BUS_DMASYNC_PREREAD:BUS_DMASYNC_PREWRITE);
 
 > ffffffff80609537:       e8 ef e6 c1 ff          callq  ffffffff80227c2b <bus_dmamap_sync>
 > ffffffff8060953c:       49 8b 76 20             mov    0x20(%r14),%rsi
 
 The crash happened repeatedly, while running either "rsync -a"
 retrieving a couple GB of remote data, or while running "git checkout"
 locally on the netbsd-src.git repository (so I/O intensive situation).
 This VM has 8 GB RAM and as much swap, running on "Joyent SmartDC HVM
 (7.20150420T105949Z)" with 2 CPUs configured.
 
 Let me know if more details are required to diagnose this.
 
 HTH,
 -- 
 khorben