Date: Mon, 28 Dec 2009 12:02:21 -0600 From: Robert Noland <rnoland@FreeBSD.org> To: "Carlos A. M. dos Santos" <unixmania@gmail.com> Cc: freebsd-x11@freebsd.org Subject: Re: x11perf -all leads to Xorg crash on 8-STABLE Message-ID: <1262023342.30049.3.camel@balrog.2hip.net> In-Reply-To: <e71790db0912280728l59a80c86rc76a3e760b5a21d7@mail.gmail.com> References: <e71790db0912261431h7285eeb4yeb1f648694730b@mail.gmail.com> <alpine.BSF.2.00.0912262126540.32530@wonkity.com> <e71790db0912270917n4ccdb185pe65f2f5c4f30ea0c@mail.gmail.com> <1261934521.2220.2834.camel@balrog.2hip.net> <e71790db0912270941hdd0bc04oe3c66a81b663f6eb@mail.gmail.com> <1261942898.2220.2976.camel@balrog.2hip.net> <e71790db0912271633q5178b868u26514a7e1ba6ec87@mail.gmail.com> <e71790db0912271735h7a3f7096rdad098e2f945243a@mail.gmail.com> <1261972205.2220.3468.camel@balrog.2hip.net> <e71790db0912280728l59a80c86rc76a3e760b5a21d7@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, 2009-12-28 at 13:28 -0200, Carlos A. M. dos Santos wrote: > On Mon, Dec 28, 2009 at 1:50 AM, Robert Noland <rnoland@freebsd.org> wrote: > > On Sun, 2009-12-27 at 23:35 -0200, Carlos A. M. dos Santos wrote: > >> I just noticed that the CONFIGURE_ARGS in x11/pixman should disable > >> sse2 on AMD64, but this is not working: > >> > >> > #0 0x000000080107c255 in pixman_blt_sse2 (src_bits=0x81344f000, > >> > dst_bits=0x804000000, src_stride=2400, > >> > dst_stride=4096, src_bpp=32, dst_bpp=32, src_x=576, src_y=576, > >> > dst_x=3, dst_y=3, width=10, height=9) > >> > at pixman-sse2.c:5330 > >> > >> I rebuilt and reinstalled the pixman with > >> > >> # cd /usr/ports/x11/pixman > >> # make BATCH=yes CONFIGURE_ARGS='--disable-vmx --disable-arm-simd > >> --disable-mmx --disable-sse2' FORCE_PKG_REGISTER=yes install > >> > >> This did not solve the problem, just gave me a different kind of crash: > > > > How about updating pixman? Patch attached. It may not be an issue in > > pixman, though. I can't tell exactly what is invalid from the trace. > > If you run it with drm debugging enabled, does drm return an error that > > is triggering the server to abort? What GPU is this on? > > Updating pixman did not help: > > Program received signal SIGSEGV, Segmentation fault. > 0x000000080107d215 in pixman_blt_sse2 (src_bits=0x81364f000, > dst_bits=0x804200000, src_stride=2400, dst_stride=4096, src_bpp=32, > dst_bpp=32, src_x=576, src_y=576, dst_x=3, dst_y=3, width=10, > height=9) > at pixman-sse2.c:5358 > 5358 *(uint32_t *)d = *(uint32_t *)s; > (gdb) print d > $26 = (uint8_t *) 0x80420300c "���" > (gdb) print s > $27 = (uint8_t *) 0x8137a1100 <Error reading address 0x8137a1100: Bad address> Ok, following the math in pixman... I don't quite see how this happens. Are you using any unusual compiler optimizations? I just tested this with pixman-0.16.0 from ports and xserver 1.7.3 with nouveau and it does not kill the server. I'm not sure that the output is correct, but it does complete without error. robert. > The GPU is Intel 945GM > > Program received signal SIGSEGV, Segmentation fault. > 0x000000080107d215 in pixman_blt_sse2 (src_bits=0x81364f000, > dst_bits=0x804200000, src_stride=2400, dst_stride=4096, src_bpp=32, > dst_bpp=32, src_x=576, src_y=576, dst_x=3, dst_y=3, width=10, > height=9) > at pixman-sse2.c:5358 > 5358 *(uint32_t *)d = *(uint32_t *)s; > (gdb) print d > $26 = (uint8_t *) 0x80420300c "���" > (gdb) print s > $27 = (uint8_t *) 0x8137a1100 <Error reading address 0x8137a1100: Bad address> > > Enabling DRM debug I get this: > > [drm:pid14907:drm_modeset_ctl] pre-modeset, crtc 1 > [drm:pid14907:drm_vblank_get] enabling vblank on crtc 1, ret: 0 > [drm:pid14907:drm_update_vblank_count] enabling vblank interrupts on > crtc 1, missed 776 > [drm:pid14907:drm_ioctl] pid=14907, cmd=0xc0086444, nr=0x44, dev > 0xffffff0001560b00, auth=1 > [drm:pid14907:i915_emit_irq] emitting: 827225 > [drm:pid14907:drm_ioctl] pid=14907, cmd=0x80046445, nr=0x45, dev > 0xffffff0001560b00, auth=1 > [drm:pid14907:drm_ioctl] pid=14907, cmd=0x2000641f, nr=0x1f, dev > 0xffffff0001560b00, auth=1 > [drm:pid14907:drm_ioctl] no function > pid 14907 (Xorg), uid 0: exited on signal 6 (core dumped) > [drm:pid14907:drm_close] open_count = 1 > [drm:pid14907:drm_close] pid = 14907, device = 0xffffff0001560b00, > open_count = 1 > [drm:pid14907:drm_close] Process 14907 dead, freeing lock for context 5 > [drm:pid14907:drm_lastclose] > > The GPU is Intel 945GM > > hostb0@pci0:0:0:0: class=0x060000 card=0x30aa103c chip=0x27a08086 > rev=0x03 hdr=0x00 > vendor = 'Intel Corporation' > device = '955XM/945GM/PM/GMS/940GML Express Processor to DRAM > Controller' > class = bridge > subclass = HOST-PCI > vgapci0@pci0:0:2:0: class=0x030000 card=0x30aa103c chip=0x27a28086 > rev=0x03 hdr=0x00 > vendor = 'Intel Corporation' > device = 'Mobile 945GM/GU Express Integrated Graphics Controller' > class = display > subclass = VGA > vgapci1@pci0:0:2:1: class=0x038000 card=0x30aa103c chip=0x27a68086 > rev=0x03 hdr=0x00 > vendor = 'Intel Corporation' > device = 'Mobile 945GM/GU Express Integrated Graphics Controller' > class = display > > > BTW, the problem does not seem to be specific to Intel GPU or FreeBSD. > I also get a X server crash by running x11perf -shmputxy10 on another > machine running Ubuntu 8.10, with a nVidia GPU: > > 00:05.0 VGA compatible controller: nVidia Corporation C51 [Geforce > 6150 Go] (rev a2) > Subsystem: Hewlett-Packard Company Device 30b7 > Flags: bus master, 66MHz, fast devsel, latency 0, IRQ 18 > Memory at b2000000 (32-bit, non-prefetchable) [size=16M] > Memory at c0000000 (64-bit, prefetchable) [size=256M] > Memory at b1000000 (64-bit, non-prefetchable) [size=16M] > [virtual] Expansion ROM at 40000000 [disabled] [size=128K] > Capabilities: <access denied> > Kernel driver in use: nvidia > Kernel modules: nvidiafb, nvidia -- Robert Noland <rnoland@FreeBSD.org> FreeBSD
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1262023342.30049.3.camel>