Re: GENERIC_SUN4U kernel trap 30: data access exception

On Fri, Oct 23, 2015 at 09:40:10AM +0200, Manuel Bouyer wrote:
> On Thu, Oct 22, 2015 at 12:34:55PM +0200, Manuel Bouyer wrote:
> > On Thu, Oct 22, 2015 at 09:19:46AM +0200, Martin Husemann wrote:
> > > On Wed, Oct 21, 2015 at 07:51:43PM +0200, Manuel Bouyer wrote:
> > > > Hello,
> > > > on a netra box running a 32bit GENERIC_SUN4U kernel (this is the reason
> > > > for the cross-post :) NetBSD 7.0, I just got:
> > > > login: cpu0: data fault: pc=141320c rpc=620401032123a47 addr=0
> > > > kernel trap 30: data access exception
> > > > Stopped in pid 21062.1 (cc1plus) at     netbsd:pmap_remove_pv+0x6c:     ld              [
> > > > %i0 + %g0], %g3
> > > 
> > > Can you reproduce it? Please file a PR!
> > 
> > No, It happended only once while running pbulk.
> > It appended while building seamonkey; after reboot the package build did
> > complete
> 
> I got another one (still running pbulk):
> cpu0: data fault: pc=15c8570 rpc=ff00080000000000 addr=0
> kernel trap 30: data access exception
> Stopped in pid 2456.1 (xsltproc) at     netbsd:amap_wipeout+0x90:       ld              [
> %g1 + 0x4], %g2
> db{0}> tr
> uvm_unmap_detach(422e718, 0, 4227dc0, 0, 1cbde40, 422e718) at netbsd:uvm_unmap_detach+0x3c
> uvmspace_free(4227d90, 3f06810, ffffffff, 1c65c80, 0, 4227d9c) at netbsd:uvmspace_free+0xd8
> exit1(41c2560, 8b, 1c9c400, 1c82400, 30a4bc00, 421fb10) at netbsd:exit1+0x154
> sigexit(41c2560, 8b, 421fb10, 180000, 8b, 0) at netbsd:sigexit+0x13c
> postsig(b, 0, 421fb10, 41c26f4, 30a4bd00, 41c2560) at netbsd:postsig+0xd4
> lwp_userret(41c2560, 1000000, 20000, 100000, 91a0002, 421fb10) at netbsd:lwp_userret+0x164
> data_access_fault(30a4bed0, c8, ff53fa0c, 1232, 0, 421fb10) at netbsd:data_access_fault+0x1ac
> ?(fce0ca10, a, 0, 24e30, 800, fdd4b000) at 10106cc

And another one:
trap type 0x34: cpu 0, pc=15dcff4 npc=15dcff8 pstate=0xffffffffff82000e<AM,PRIV,IE>
kernel trap 34: mem address not aligned
Stopped in pid 0.37 (system) at netbsd:uvmpd_trylockowner+0x14: ld              [ %g1 + %g0], %i0
db{0}> tr
uvmpdpol_balancequeue(0, 4bdcd00, 0, 673c, 38b25d0, 31509e0) at netbsd:uvmpdpol_balancequeue+0x5c
uvm_pageout(0, 1f242, 0, 0, 0, 1cbd44c) at netbsd:uvm_pageout+0x290
lwp_trampoline(f0075db8, fffa3cf8, 111800, 1106c8, fffa3df8, 0) at netbsd:lwp_trampoline+0x8

It looks like memroy corruption ...

-- 
Manuel Bouyer <[email protected]>
     NetBSD: 26 ans d'experience feront toujours la difference
--