Stefan Weil [Tue, 29 Apr 2014 06:21:16 +0000 (08:21 +0200)]
configure: Enable dead code (lzo, snappy, quorum)
Those options were not enabled by default, even when the build
environment would have supported them, so the corresponding
code was not compiled in normal test builds like on build bots.
[Building quorum by default "broke" qemu-iotests ./check 081. It turns
out the 081.out master output was just bitrotted. Fix this by updating
the error message.
--Stefan]
David Marchand [Wed, 11 Jun 2014 15:25:16 +0000 (17:25 +0200)]
char: fix avail_connections init in qemu_chr_open_eventfd()
When trying to use a ivshmem server with qemu, ivshmem init code tries to
create a CharDriverState object for each eventfd retrieved from the server.
To create this object, a call to qemu_chr_open_eventfd() is done.
Right after this, before adding a frontend, qemu_chr_fe_claim_no_fail() is
called.
qemu_chr_open_eventfd() does not set avail_connections to 1, so no frontend can
be associated because qemu_chr_fe_claim_no_fail() makes qemu stop right away.
Peter Maydell [Thu, 12 Jun 2014 08:51:41 +0000 (09:51 +0100)]
Merge remote-tracking branch 'remotes/kraxel/tags/pull-gtk-20140611-1' into staging
gtk: misc fixes & cleanups.
# gpg: Signature made Wed 11 Jun 2014 13:28:12 BST using RSA key ID D3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <[email protected]>"
# gpg: aka "Gerd Hoffmann <[email protected]>"
# gpg: aka "Gerd Hoffmann (private) <[email protected]>"
* remotes/kraxel/tags/pull-gtk-20140611-1:
gtk: update window size after showing/hiding tabs
gtk: factor out gtk3 grab into the new gd_grab_devices function
gtk: cleanup backend dependencies
gtk: factor out keycode mapping
Peter Maydell [Wed, 11 Jun 2014 17:05:21 +0000 (18:05 +0100)]
Merge remote-tracking branch 'remotes/pmaydell/tags/pull-bsd-user-20140611' into staging
bsd-user queue:
* build fixes
* improvements to strace
# gpg: Signature made Wed 11 Jun 2014 15:23:40 BST using RSA key ID 14360CDE
# gpg: Good signature from "Peter Maydell <[email protected]>"
* remotes/pmaydell/tags/pull-bsd-user-20140611:
bsd-user: Fix syscall format, add strace support for more syscalls
bsd-user: Implement strace support for thr_* syscalls
bsd-user: Implement strace support for extattr_* syscalls
bsd-user: Implement strace support for __acl_* syscalls
bsd-user: Implement strace support for print_ioctl syscall
bsd-user: Implement strace support for print_sysctl syscall
bsd-user: GPL v2 attribution update and style
bsd-user: add HOST_VARIANT_DIR for various *BSD dependent code
exec: replace ffsl with ctzl
vhost: replace ffsl with ctzl
xen: replace ffsl with ctzl
util/qemu-openpty: fix build with musl libc by include termios.h as fallback
bsd-user/mmap.c: Don't try to override g_malloc/g_free
util/hbitmap.c: Use ctpopl rather than reimplementing a local equivalent
bsd-user: refresh freebsd system call numbers
Peter Maydell [Wed, 11 Jun 2014 14:36:48 +0000 (15:36 +0100)]
Merge remote-tracking branch 'remotes/bonzini/configure' into staging
* remotes/bonzini/configure:
rules.mak: Rewrite unnest-vars
configure: unset interfering variables
configure: duplicate/incorrect order of -lrt
libcacard: improve documentation
libcacard: actually use symbols file
libcacard: replace qemu thread primitives with glib ones
vscclient: use glib thread primitives not qemu
glib-compat.h: add new thread API emulation on top of pre-2.31 API
Laszlo Ersek [Tue, 20 May 2014 11:42:53 +0000 (13:42 +0200)]
dump: simplify get_len_buf_out()
We can (and should) rely on the fact that s->flag_compress is exactly one
of DUMP_DH_COMPRESSED_ZLIB, DUMP_DH_COMPRESSED_LZO, and
DUMP_DH_COMPRESSED_SNAPPY.
This is ensured by the QMP schema and dump_init() in combination.
This patch doesn't change the fact that lzo_init() is called for every
LZO-compressed dump, but it makes get_len_buf_out() more focused (single
responsibility).
Laszlo Ersek [Tue, 20 May 2014 11:39:45 +0000 (13:39 +0200)]
dump: select header bitness based on ELF class, not ELF architecture
The specific ELF architecture (d_machine) carries Too Much Information
(TM) for deciding between create_header32() and create_header64(), use
"d_class" instead (ELFCLASS32 vs. ELFCLASS64).
This change adapts write_dump_header() to write_elf_loads(), dump_begin()
etc. that also rely on the ELF class of the target for bitness selection.
Considering the current targets that support dumping, cpu_get_dump_info()
works as follows:
- target-s390x/arch_dump.c: (EM_S390, ELFCLASS64) only
- target-ppc/arch_dump.c (EM_PPC64, ELFCLASS64) only
- target-i386/arch_dump.c: sets (EM_X86_64, ELFCLASS64) vs. (EM_386,
ELFCLASS32) keying off the same Long Mode Active flag.
Use TARGET_PAGE_SIZE and ~TARGET_PAGE_MASK instead.
"DumpState.page_size" has type "size_t", whereas TARGET_PAGE_SIZE has type
"int". TARGET_PAGE_MASK is of type "int" and has negative value. The patch
affects the implicit type conversions as follows:
- create_header32() and create_header64(): assigned to "block_size", which
has type "uint32_t". No change.
- get_next_page(): "block->target_start", "block->target_end" and "addr"
have type "hwaddr" (uint64_t).
Before the patch,
- if "size_t" was "uint64_t", then no additional conversion was done as
part of the usual arithmetic conversions,
- If "size_t" was "uint32_t", then it was widened to uint64_t as part of
the usual arithmetic conversions,
for the remainder and addition operators.
After the patch,
- "~TARGET_PAGE_MASK" expands to ~~((1 << TARGET_PAGE_BITS) - 1). It
has type "int" and positive value (only least significant bits set).
That's converted (widened) to "uint64_t" for the bit-ands. No visible
change.
- The same holds for the (addr + TARGET_PAGE_SIZE) addition.
- write_dump_pages():
- TARGET_PAGE_SIZE passed as argument to a bunch of functions that all
have prototypes. No change.
- When incrementing "offset_data" (of type "off_t"): given that we never
build for ILP32_OFF32 (see "-D_FILE_OFFSET_BITS=64" in configure),
"off_t" is always "int64_t", and we only need to consider:
- ILP32_OFFBIG: "size_t" is "uint32_t".
- before: int64_t += uint32_t. Page size converted to int64_t for
the addition.
- after: int64_t += int32_t. No change.
- LP64_OFF64: "size_t" is "uint64_t".
- before: int64_t += uint64_t. Offset converted to uint64_t for the
addition, then the uint64_t result is converted to int64_t for
storage.
- after: int64_t += int32_t. Same as the ILP32_OFFBIG/after case.
No visible change.
- (size_out < s->page_size) comparisons, and (size_out = s->page_size)
assignment:
- before: "size_out" is of type "size_t", no implicit conversion for
either operator.
- after: TARGET_PAGE_SIZE (of type "int" and positive value) is
converted to "size_t" (for the relop because the latter is
one of "uint32_t" and "uint64_t"). No visible change.
- dump_init():
- DIV_ROUND_UP(DIV_ROUND_UP(s->max_mapnr, CHAR_BIT), s->page_size): The
innermost "DumpState.max_mapnr" field has type uint64_t, which
propagates through all implicit conversions at hand:
#define DIV_ROUND_UP(n,d) (((n) + (d) - 1) / (d))
regardless of the page size macro argument's type. In the outer macro
replacement, the page size is converted from uint32_t and int32_t
alike to uint64_t.
- (tmp * s->page_size) multiplication: "tmp" has size "uint64_t"; the
RHS is converted to that type from uint32_t and int32_t just the same
if it's not uint64_t to begin with.
"DumpState.page_shift" used to have type "uint32_t", while the replacement
TARGET_PAGE_BITS has type "int". Since "DumpState.page_shift" was only
used as bit shift counts in the paddr_to_pfn() and pfn_to_paddr() macros,
this is safe.
Laszlo Ersek [Tue, 20 May 2014 11:39:42 +0000 (13:39 +0200)]
dump: simplify write_start_flat_header()
Currently, the function
- defines and populates an auto variable of type MakedumpfileHeader
- allocates and zeroes a buffer of size MAX_SIZE_MDF_HEADER (4096)
- copies the former into the latter (covering an initial portion of the
latter)
Fill in the MakedumpfileHeader structure in its final place (the alignment
is OK because the structure lives at the address returned by g_malloc0()).
Laszlo Ersek [Tue, 20 May 2014 11:39:41 +0000 (13:39 +0200)]
dump: fill in the flat header signature more pleasingly to the eye
The "mh.signature" array field has size 16, and is zeroed by the preceding
memset(). MAKEDUMPFILE_SIGNATURE expands to a string literal with string
length 12 (size 13). There's no need to measure the length of
MAKEDUMPFILE_SIGNATURE at runtime, nor for the extra zero-filling of
"mh.signature" with strncpy().
Use memcpy() with MIN(sizeof, sizeof) for robustness (which is an integer
constant expression, evaluable at compile time.)
Gerd Hoffmann [Wed, 28 May 2014 20:33:06 +0000 (22:33 +0200)]
gtk: cleanup backend dependencies
Make configure detect gtk x11 backend and link libX11 then. Make
gtk backend specific code properly #ifdef'ed on the GTK_WINDOWING_*
backends at runtime). Our gtk ui code should build and run fine on
any platform now.
Natanael Copa [Tue, 29 Apr 2014 14:17:27 +0000 (16:17 +0200)]
util/qemu-openpty: fix build with musl libc by include termios.h as fallback
Include termios.h as POSIX fallback when not glibc, bsd or solaris.
POSIX says that termios.h should define struct termios and TCAFLUSH.
http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/termios.h.html
This fixes the following compile errors with musl libc:
util/qemu-openpty.c: In function 'qemu_openpty_raw':
util/qemu-openpty.c:112:20: error: storage size of 'tty' isn't known
struct termios tty;
^
...
util/qemu-openpty.c:128:24: error: 'TCSAFLUSH' undeclared (first use in this function)
tcsetattr(*aslave, TCSAFLUSH, &tty);
^
Peter Maydell [Tue, 3 Jun 2014 23:13:31 +0000 (00:13 +0100)]
bsd-user/mmap.c: Don't try to override g_malloc/g_free
Trying to override the implementations of g_malloc and g_free is
a really bad idea -- it means statically linked builds fail to
link (because of the multiple definitions provided by this file
and by glib), and non-statically linked builds segfault as soon
as they try to do anything more complicated than printing the
usage message. Remove these overridden versions and just use
the glib ones.
This is sufficient that bsd-user can run basic x86-64
binaries on OpenBSD again; FreeBSD and NetBSD seem to have
further issues.
Peter Maydell [Tue, 3 Jun 2014 23:07:20 +0000 (00:07 +0100)]
util/hbitmap.c: Use ctpopl rather than reimplementing a local equivalent
The function popcountl() in hbitmap.c is effectively a reimplementation
of what host-utils.h provides as ctpopl(). Use ctpopl() directly; this fixes
a failure to compile on NetBSD (whose strings.h erroneously exposes a
system popcountl() which clashes with this one).
Peter Maydell [Tue, 10 Jun 2014 16:16:03 +0000 (17:16 +0100)]
Merge remote-tracking branch 'remotes/mjt/tags/trivial-patches-2014-06-10' into staging
trivial patches for 2014-06-10
# gpg: Signature made Tue 10 Jun 2014 17:07:19 BST using RSA key ID A4C3D7DB
# gpg: Good signature from "Michael Tokarev <[email protected]>"
# gpg: aka "Michael Tokarev <[email protected]>"
# gpg: aka "Michael Tokarev <[email protected]>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 6EE1 95D1 886E 8FFB 810D 4324 457C E0A0 8044 65C5
# Subkey fingerprint: 6F67 E18E 7C91 C5B1 5514 66A7 BEE5 9D74 A4C3 D7DB
* remotes/mjt/tags/trivial-patches-2014-06-10: (25 commits)
virtio.c: fix error message
hw: vmware_vga: don't return cursorx when the driver asks for cursory register
migration: Plug memory leak in migrate-set-cache-size command
libcacard: Clean up dead stores before g_free()
libcacard: Drop superfluous conditionals around g_free()
cpu/x86: correctly set errors in x86_cpu_parse_featurestr
smbios: use g_free directly on NULL pointers
vdi: remove double conversion
apb: Fix compiler warnings (large constants)
hw/net/ne2000-isa: Register vmstate struct
target-microblaze: Delete unused sign_extend() function
hw/misc/milkymist-softusb: Remove unused softusb_{read, write}_pmem()
target-i386/translate.c: Remove unused tcg_gen_lshift()
hw/isa/pc87312: Remove unused function is_parallel_epp()
hw/intc/openpic: Remove unused function IRQ_testbit()
hw/dma/xilinx_axidma: Remove unused stream_halted() function
util/qemu-sockets.c: Avoid unused variable warnings
hw/sd/sd.c: Drop unused sd_acmd_type[] array
hw/i386/pc.c: Remove unused parallel_io and parallel_irq variables
slirp: Remove unused zero_ethaddr[] variable
...
The 'on' variable is never used, and 'off' is only used
if IPV6_V6ONLY is defined; delete 'on' and move 'off' to
the point where it is used. This avoids warnings from
clang 3.4.
Peter Maydell [Tue, 3 Jun 2014 17:29:01 +0000 (18:29 +0100)]
hw/sd/sd.c: Drop unused sd_acmd_type[] array
Drop the sd_acmd_type[] array: it is never used. (The equivalent
sd_cmd_type[] array for normal commands is used to identify
those commands whose argument includes the card address in the
top 16 bits; but for app commands the card address is passed
with the APP_CMD prefix, not with the argument to the app command
itself.)
Add a debug printf for TX descriptor fetching. This is helpful to anyone
needing to debug TX ring buffer traversal. It is also now consistent with
the RX code which has a similar printf.
The local variable "desc" was being used to read-modify-write the
first descriptor (of a multi-desc packet) upon packet completion.
desc however continues to be used by the code as the current
descriptor. Give this first desc RMW it's own local variable to
avoid trampling.
Fam Zheng [Tue, 27 May 2014 07:54:19 +0000 (15:54 +0800)]
rules.mak: Rewrite unnest-vars
The macro unnest-vars is the most important, complicated but hard to
track magic in QEMU's build system.
Rewrite it in a (hopefully) clearer way, with more comments, to make it
easier to understand and maintain.
Remove DSO_CFLAGS and module-objs-m that are not used.
A bonus fix of this version is, per object variables are properly
protected in save-objs and load-objs, before including sub-dir
Makefile.objs, just as nested variables are. So the occasional same
object name from different directory levels won't step on each other's
foot.
Cornelia Huck [Wed, 28 May 2014 10:39:17 +0000 (12:39 +0200)]
configure: unset interfering variables
The check for big or little endianness relies on grep reporting
match/non-match on the generated binary. If the user specified
--binary-files=without-match in their GREP_OPTIONS, this will fail.
Let's follow what autoconf does and unset GREP_OPTIONS and CLICOLOR_FORCE
at the beginning of the script.
Peter Maydell [Tue, 10 Jun 2014 09:59:26 +0000 (10:59 +0100)]
Merge remote-tracking branch 'remotes/cohuck/tags/s390x-20140610' into staging
Several patches for s390:
- bugfixes: A fix for a long-standing bug in the css code as well as
a fixup for the recent I/O adapter support.
- Exploitation of the userspace cmma enablement/reset interface, if
it is present.
- Some debuggability improvements by logging unmanageable conditions.
- virtio-ccw finally gets migration support for its structures.
- Some cleanup as to how floating interrupts are injected.
# gpg: Signature made Tue 10 Jun 2014 08:57:56 BST using RSA key ID C6F02FAF
# gpg: Can't check signature: public key not found
* remotes/cohuck/tags/s390x-20140610:
s390x/kvm: inject via flic
s390x: cleanup interrupt injection
s390x/kvm: add alternative injection interface
s390x: consolidate floating interrupts
s390/virtio-ccw: migration support
s390x/kvm: Log unmanageable program interruptions
s390x/kvm: Log unmanageable external interruptions
s390x/kvm: enable/reset cmma via vm attributes
s390x/kvm: make flic play well with old kernels
s390x/css: handle emw correctly for tsch
Thomas Huth [Wed, 7 May 2014 07:45:21 +0000 (09:45 +0200)]
s390x/kvm: Log unmanageable program interruptions
The kernel only drops to userspace if an endless program interrupt loop
has been detected. Let's print an error message in this case to inform
the user about the crash and stop the affected CPU with a panic event,
just like it is already done for the external interruption loop detection.
Interception code 0x14 only drops to userspace when an unmanageable
external interruption interception occured (e.g. if the External New
PSW does not disable external interruptions). Instead of bailing out
via the default handler, it is better to inform the user with a
proper error message that also includes the bad PSW, and to stop
the affected CPU with a panic event instead.
Exploit the new api for userspace-controlled cmma. If supported, enable
cmma during kvm initialization and register a reset handler for cmma,
which is also called directly from the load IPL code.
The reset functionality is needed to reset the cmma state of the guest
pages, e.g. if a system reset is triggered via qemu monitor; otherwise
this could result in data corruption.
A guest triggered reboot may now lead to multiple cmma resets; this is
OK, however, as this is slowpath anyway and the simplest way to achieve
the intended effects.
Cornelia Huck [Wed, 28 May 2014 12:15:57 +0000 (14:15 +0200)]
s390x/kvm: make flic play well with old kernels
If we run with an old kernel that does not support KVM_CAP_IRQ_ROUTING,
we don't have to do anything in the ->register_io_adapter and
->io_adapter_map callbacks and therefore should return 0 instead of
-ENOSYS (just as the non-kvm flic does).
This fixes using adapter interrupts when running under an older kernel,
which broke with "s390x: add I/O adapter registration".
Cornelia Huck [Tue, 27 May 2014 10:40:44 +0000 (12:40 +0200)]
s390x/css: handle emw correctly for tsch
We should not try to store the emw portion of the irb if extended
measurements are not applicable. In particular, we should not surprise
the guest by storing a larger irb if it did not enable extended
measurements.
Paolo Bonzini [Wed, 28 May 2014 16:13:41 +0000 (18:13 +0200)]
libcacard: improve documentation
Using the file-backed smartcard backend is black magic, but it can
be useful if your only smartcard bricks itself if it is accessed
the wrong way too many times.
Complete the documentation to include the art of creating certificates
and using them with QEMU, based on Ray Strode's useful tutorial at
https://blogs.gnome.org/halfline/2013/09/08/another-smartcard-post/
but with ccid-card-emulated or vscclient instead of SPICE.
Michael Tokarev [Thu, 8 May 2014 08:30:49 +0000 (12:30 +0400)]
libcacard: actually use symbols file
libtool has an argument for .syms file, which is -export-symbols.
There's no argument `-export-syms', and it looks like at least on
linux, -export-syms is just ignored. Use the correct argument,
-export-symbols, to actually get the right export list.
Michael Tokarev [Thu, 8 May 2014 08:30:47 +0000 (12:30 +0400)]
vscclient: use glib thread primitives not qemu
Use glib-provided thread primitives in vscclient instead of
qemu ones, and do not use qemu sockets in there (open-code
call to WSAStartup() for windows to initialize things).
This way, vscclient becomes more stand-alone, independent on
qemu internals.
Michael Tokarev [Thu, 8 May 2014 08:30:46 +0000 (12:30 +0400)]
glib-compat.h: add new thread API emulation on top of pre-2.31 API
Thread API changed in glib-2.31 significantly. Before that version,
conditionals and mutexes were only allocated dynamically, using
_new()/_free() interface. in 2.31 and up, they're allocated statically
as regular variables, and old interface is deprecated.
(Note: glib docs says the new interface is available since version
2.32, but it was actually introduced in version 2.31).
Create the new interface using old primitives, by providing non-opaque
definitions of the base types (GCond and GMutex) using GOnces.
Replace #ifdeffery around GCond and GMutex in trace/simple.c and
coroutine-gthread.c too because it does not work anymore with the new
glib-compat.h.
Signed-off-by: Michael Tokarev <[email protected]> Reviewed-by: Stefan Hajnoczi <[email protected]>
[Use GOnce to support lazy initialization; introduce CompatGMutex
and CompatGCond. - Paolo] Signed-off-by: Paolo Bonzini <[email protected]>
projects / qemu.git / log