Symptom:
$ qemu-system-x86_64 -m 10000000
Unexpected error in ram_block_add() at /work/armbru/qemu/exec.c:1456:
upstream-qemu: cannot set up guest memory 'pc.ram': Cannot allocate memory
Aborted (core dumped)
Root cause: commit ef701d7 screwed up handling of out-of-memory
conditions. Before the commit, we report the error and exit(1), in
one place, ram_block_add(). The commit lifts the error handling up
the call chain some, to three places. Fine. Except it uses
&error_abort in these places, changing the behavior from exit(1) to
abort(), and thus undoing the work of commit 3922825 "exec: Don't
abort when we can't allocate guest memory".
The three places are:
* memory_region_init_ram()
Commit 4994653 (right after commit ef701d7) lifted the error
handling further, through memory_region_init_ram(), multiplying the
incorrect use of &error_abort. Later on, imitation of existing
(bad) code may have created more.
* memory_region_init_ram_ptr()
The &error_abort is still there.
* memory_region_init_rom_device()
Doesn't need fixing, because commit 33e0eb5 (soon after commit
ef701d7) lifted the error handling further, and in the process
changed it from &error_abort to passing it up the call chain.
Correct, because the callers are realize() methods.
Fix the error handling after memory_region_init_ram() with a
Coccinelle semantic patch:
@r@
expression mr, owner, name, size, err;
position p;
@@
memory_region_init_ram(mr, owner, name, size,
(
- &error_abort
+ &error_fatal
|
err@p
)
);
@script:python@
p << r.p;
@@
print "%s:%s:%s" % (p[0].file, p[0].line, p[0].column)
When the last argument is &error_abort, it gets replaced by
&error_fatal. This is the fix.
If the last argument is anything else, its position is reported. This
lets us check the fix is complete. Four positions get reported:
* ram_backend_memory_alloc()
Error is passed up the call chain, ultimately through
user_creatable_complete(). As far as I can tell, it's callers all
handle the error sanely.
* fsl_imx25_realize(), fsl_imx31_realize(), dp8393x_realize()
DeviceClass.realize() methods, errors handled sanely further up the
call chain.
We're good. Test case again behaves:
$ qemu-system-x86_64 -m 10000000
qemu-system-x86_64: cannot set up guest memory 'pc.ram': Cannot allocate memory
[Exit 1 ]
The next commits will repair the rest of commit ef701d7's damage.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <1441983105-26376-3-git-send-email-armbru@redhat.com>
Reviewed-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
We create optional sections with this patch. But we already have
optional subsections. Instead of having two mechanism that do the
same, we can just generalize it.
For subsections we just change:
- Add a needed function to VMStateDescription
- Remove VMStateSubsection (after removal of the needed function
it is just a VMStateDescription)
- Adjust the whole tree, moving the needed function to the corresponding
VMStateDescription
Signed-off-by: Juan Quintela <quintela@redhat.com>
Make the code a bit more obvious.
We don't have min/max, so a general helper for clamp probably isn't
acceptable either.
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
vga_common_init() doesn't allow more than 256 MiB vram size and silently
shrinks any larger value. qxl_dirty_surfaces() used the unshrinked size
via qxl->shadow_rom.surface0_area_size when accessing the memory, which
resulted in segfault.
Add a workaround for this case and an assert if it happens again.
We have to bump the vga memory limit too, because 256 MiB wouldn't have
allowed 8k (it requires more than 128 MiB).
1024 MiB doesn't work, but 512 MiB seems fine.
Proposed-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Now that isa_mem_base variable is always 0, we can remove its usage.
Signed-off-by: Hervé Poussineau <hpoussin@reactos.org>
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
Warning from the Sparse static analysis tool:
hw/display/vga.c:2012:26: warning:
symbol 'vmstate_vga_endian' was not declared. Should it be static?
Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
This allows VGA to decide whether to use a shared surface based on
whether the UI backend supports the format or not. Backends that
don't provide the new callback fallback to native 32 bpp which
is equivalent to what was supported before.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[ kraxel: fix console check, allow only 32 bpp as fallback ]
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This prevents surface sharing which will be necessary to
fix cirrus HW cursor support.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Following cleanup of the vga device code in commit d2e043a804,
the arrays dmask4 and dmask16 are now unused. gcc doesn't warn
about this, but clang does; remove them.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Include the endian state in the migration stream as an optional
subsection which we only include when the endian isn't the default,
thus enabling backward compatibility of the common case.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Changes by kraxel:
* Remove bochs dispi interface changes. We'll do that in
a different way to make sure we don't conflict with
possible future bochs dispi interface changes.
* keep live migration bits.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
And initialize it based on target endian
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
It's no longer a template, we only instanciate the file once.
Keep it a #included file so the functions remain static.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Not all platforms have a VGA BIOS, powerpc typically relies on
using the DISPI interface to initialize the card.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
The test for surface_bits_per_pixel() isn't necessary anymore,
the 8bpp case never happens.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
We always use rgb_to_pixel32 nowadays.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Provide different functions for converting from an LE vs a BE
framebuffer. We cannot rely on the simple cases always being
shared surfaces since cirrus will need to always shadow for
cursor emulation, so we need the full set of functions to
be able to later handle runtime switching.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>\
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
All the macros used to generate different versions of vga_template.h
are now unnecessary, take them all out and remove the _32 suffix from
most functions.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Nowadays, we either share a surface with the host, or we create
a 32bpp ARGB console surface.
So we only need to draw/convert to 32bpp, enabling us to remove
all but one instance of vga_template.h inclusion (to be further
cleaned up), rgb_to_pixel_* etc...
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
- Build: fixing block/iscsi.so and ranlib warnings on Mac OS X
- Migration fixes for x86
- The odd KVM patch.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2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=GIKW
-----END PGP SIGNATURE-----
Merge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging
- Memory: improve error reporting and avoid crashes on hotplug
- Build: fixing block/iscsi.so and ranlib warnings on Mac OS X
- Migration fixes for x86
- The odd KVM patch.
# gpg: Signature made Thu 11 Sep 2014 11:21:10 BST using RSA key ID 9B4D86F2
# gpg: Good signature from "Paolo Bonzini <pbonzini@redhat.com>"
# gpg: aka "Paolo Bonzini <bonzini@gnu.org>"
* remotes/bonzini/tags/for-upstream: (21 commits)
gdbstub: init mon_chr through qemu_chr_alloc
pckbd: adding new fields to vmstate
mc146818rtc: add missed field to vmstate
piix: do not set irq while loading vmstate
serial: fixing vmstate for save/restore
parallel: adding vmstate for save/restore
fdc: adding vmstate for save/restore
cpu: init vmstate for ticks and clock offset
apic_common: vapic_paddr synchronization fix
vl: use QLIST_FOREACH_SAFE to visit change state handlers
exec: add parameter errp to gethugepagesize
exec: report error when memory < hpagesize
hostmem-ram: don't exit qemu if size of memory-backend-ram is way too big
memory: add parameter errp to memory_region_init_rom_device
memory: add parameter errp to memory_region_init_ram
exec: add parameter errp to qemu_ram_alloc and qemu_ram_alloc_from_ptr
rules.mak: Fix DSO build by pulling in archive symbols
util: Don't link host-utils.o if it's empty
util: Move general qemu_getauxval to util/getauxval.c
trace: Only link generated-tracers.o with "simple" backend
...
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Add parameter errp to memory_region_init_ram and update all call sites
to pass in &error_abort.
Signed-off-by: Hu Tao <hutao@cn.fujitsu.com>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
With this patch the qemu console core stops using PixelFormat and pixman
format codes side-by-side, pixman format code is the primary way to
specify the DisplaySurface format:
* DisplaySurface stops carrying a PixelFormat field.
* qemu_create_displaysurface_from() expects a pixman format now.
Functions to convert PixelFormat to pixman_format_code_t (and back)
exist for those who still use PixelFormat. As PixelFormat allows
easy access to masks and shifts it will probably continue to exist.
[ xenfb added by Benjamin Herrenschmidt ]
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Plug a bunch of holes in the bochs dispi interface parameter checking.
Add a function doing verification on all registers. Call that
unconditionally on every register write. That way we should catch
everything, even changing one register affecting the valid range of
another register.
Some of the holes have been added by commit
e9c6149f6a. Before that commit the
maximum possible framebuffer (VBE_DISPI_MAX_XRES * VBE_DISPI_MAX_YRES *
32 bpp) has been smaller than the qemu vga memory (8MB) and the checking
for VBE_DISPI_MAX_XRES + VBE_DISPI_MAX_YRES + VBE_DISPI_MAX_BPP was ok.
Some of the holes have been there forever, such as
VBE_DISPI_INDEX_X_OFFSET and VBE_DISPI_INDEX_Y_OFFSET register writes
lacking any verification.
Security impact:
(1) Guest can make the ui (gtk/vnc/...) use memory rages outside the vga
frame buffer as source -> host memory leak. Memory isn't leaked to
the guest but to the vnc client though.
(2) Qemu will segfault in case the memory range happens to include
unmapped areas -> Guest can DoS itself.
The guest can not modify host memory, so I don't think this can be used
by the guest to escape.
CVE-2014-3615
Cc: qemu-stable@nongnu.org
Cc: secalert@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
VgaState->vram_size is the size of the pci bar. In case of qxl not the
whole pci bar can be used as vga framebuffer. Add a new variable
vbe_size to handle that case. By default (if unset) it equals
vram_size, but qxl can set vbe_size to something else.
This makes sure VBE_DISPI_INDEX_VIDEO_MEMORY_64K returns correct results
and sanity checks are done with the correct size too.
Cc: qemu-stable@nongnu.org
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Explicitly call object_unparent in the few places where we
will re-create the memory region. If the memory region is
simply being destroyed as part of device teardown, let QOM
handle it.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Instead, add a boolean variable to indicate the presence of the region.
This avoids a repeated malloc/free (later we can also avoid the
add_child/unparent by changing the offset/size of the alias).
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
After previous Peter patch, they are redundant. This way we don't
assign them except when needed. Once there, there were lots of case
where the ".fields" indentation was wrong:
.fields = (VMStateField []) {
and
.fields = (VMStateField []) {
Change all the combinations to:
.fields = (VMStateField[]){
The biggest problem (appart from aesthetics) was that checkpatch complained
when we copy&pasted the code from one place to another.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Acked-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
PortioList is an abstraction used for construction of MemoryRegionPortioList
from MemoryRegionPortio. It can be used later to unmap created memory regions.
It also requires proper cleanup because some of the memory inside is allocated
dynamically.
By moving PortioList ot device state we make it possible to cleanup later and
avoid leaking memory.
This change spans several target platforms. The following testcases cover all
changed lines:
qemu-system-ppc -M prep
qemu-system-i386 -vga qxl
qemu-system-i386 -M isapc -soundhw adlib -device ib700,id=watchdog0,bus=isa.0
Signed-off-by: Kirill Batuzov <batuzovk@ispras.ru>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
Add a standard vga variant which doesn't occupy any legacy
resources and thus can easily be used as secondary (or legacy-free)
graphics adapter. Programming must be done using the MMIO bar.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Need a way to opt-out from vga.vram being global vmstate, for
secondary vga cards. Add a bool parameter to vga_common_init
to support this.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
After calling dpy_gfx_replace_surface(s->con, surface), the outer
surface is invalid.
==5370== Invalid read of size 4
==5370== at 0x460229: surface_bits_per_pixel (console.h:250)
==5370== by 0x466A81: get_depth_index (vga.c:1173)
==5370== by 0x467EC2: vga_draw_graphic (vga.c:1718)
==5370== by 0x4687A5: vga_update_display (vga.c:1914)
==5370== by 0x2A782E: qxl_hw_update (qxl.c:1766)
==5370== by 0x3EB83B: graphic_hw_update (console.c:254)
==5370== by 0x3FBE31: qemu_spice_display_refresh (spice-display.c:418)
==5370== by 0x2A7D01: display_refresh (qxl.c:1886)
==5370== by 0x3EEE1C: dpy_refresh (console.c:1436)
==5370== by 0x3EB543: gui_update (console.c:192)
==5370== by 0x3C43B3: timerlist_run_timers (qemu-timer.c:488)
==5370== by 0x3C4416: qemu_clock_run_timers (qemu-timer.c:499)
==5370== Address 0x22ffb1e0 is 0 bytes inside a block of size 56 free'd
==5370== at 0x4A074C4: free (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so)
==5370== by 0x4245FC: free_and_trace (vl.c:2771)
==5370== by 0x50899AE: g_free (gmem.c:252)
==5370== by 0x3EE8D3: qemu_free_displaysurface (console.c:1332)
==5370== by 0x3EEDB7: dpy_gfx_replace_surface (console.c:1427)
==5370== by 0x467EB6: vga_draw_graphic (vga.c:1714)
==5370== by 0x4687A5: vga_update_display (vga.c:1914)
==5370== by 0x2A782E: qxl_hw_update (qxl.c:1766)
==5370== by 0x3EB83B: graphic_hw_update (console.c:254)
==5370== by 0x3FBE31: qemu_spice_display_refresh (spice-display.c:418)
==5370== by 0x2A7D01: display_refresh (qxl.c:1886)
==5370== by 0x3EEE1C: dpy_refresh (console.c:1436)
Signed-off-by: Marc-André Lureau <marcandre.lureau@gmail.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Message-id: 1383664554-15248-1-git-send-email-marcandre.lureau@gmail.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>
This allows to remove the explicit qemu_flush_coalesced_mmio_buffer
calls.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This is an autogenerated patch using scripts/switch-timer-api.
Switch the entire code base to using the new timer API.
Note this patch may introduce some line length issues.
Signed-off-by: Alex Bligh <alex@alex.org.uk>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Pass a single GraphicHwOps struct pointer to graphic_console_init,
instead of a bunch of function pointers.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Screendumps are alot simpler as we can update non-active
QemuConsoles now. So we only need to update the QemuConsole
we want write out, then dump the DisplaySurface content into
a ppm file. Done.
No console switching needed. No special support code in the
gfx card emulation needed. Zap it all. Also move ppm_save
out of the vga code and next to the qmp_screendump function.
For now screen dumping is limited to console #0 (like it used
to be), even though it is dead simple to extend it to other
consoles. I wanna finish the console cleanup before setting
new qapi interfaces into stone.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Tested-by: Igor Mitsyanko <i.mitsyanko@gmail.com>
Add QemuConsole parameter to vga_hw_*, so the interface allows to update
non-active consoles (the actual code can't handle this yet, see next
patch). Passing NULL is allowed and updates the active console, like
the functions do today.
While touching all vga_hw_* calls anyway rename that to the functions to
hardware-neutral graphics_hw_*
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
