Byte swap was applied in the wrong order with testing for
RLIM_INFINITY. On mips bigendian from an amd64 system this results in
infinity being misinterpretted as 2^31-1.
This is a serious bug because it causes setrlimit stack size to kill
all child processes. This means (for example) that 'make' can run no
children. The mechanism of failure:
1. parent sets stack size rlimit to 'infinity'
2. qemu screws this value up
3. child process fetches stack size as a large (but non-infinite) value
4. qemu tries to allocate stack before execution
5. stack allocation fails (too big) and child process dies
Signed-off-by: Wesley W. Terpstra <terpstra@debian.org>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
Enforce the same restriction on the size of the sigset passed to
pselect6 as the Linux kernel does. This is both correct and silences
a gcc 4.6 warning about a write-only variable.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
As noticed while looking at "Bump do_syscall() up to 8 syscall arguments"
patch, sync_file_range uses a pad argument on 32bit mips. Deal with it
by reading the correct arguments when on mips.
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
On 32 bit MIPS a few syscalls have 7 arguments, and so to call
them via NR_syscall the guest needs to be able to pass 8 arguments
to do_syscall(). Raise the number of arguments do_syscall() takes
accordingly.
This fixes some gcc 4.6 compiler warnings about arg7 and arg8
variables being set and never used.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
Looking at the other architectures, we should be using "how" not "arg1".
Signed-off-by: Juan Quintela <quintela@redhat.com>
[peter.maydell@linaro.org: remove unnecessary initialisation of how]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
We assign ret with the error code, but then return 0 unconditionally.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
Some architectures (like Blackfin) only implement pselect6 (and skip
select/newselect). So add support for it.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
There were several remaining bugs in the previous implementation of
do_brk():
1. the value of "new_alloc_size" was one page too large when the
requested brk was aligned on a host page boundary.
2. no new pages should be (re-)allocated when the requested brk is
in the range of the pages that were already allocated
previsouly (for the same purpose). Technically these pages are
never unmapped in the current implementation.
The problem/fix can be reproduced/validated with the test-suite above:
#include <unistd.h> /* syscall(2), */
#include <sys/syscall.h> /* SYS_brk, */
#include <stdio.h> /* puts(3), */
#include <stdlib.h> /* exit(3), EXIT_*, */
#include <stdint.h> /* uint*_t, */
#include <sys/mman.h> /* mmap(2), MAP_*, */
#include <string.h> /* memset(3), */
int main()
{
int exit_status = EXIT_SUCCESS;
uint8_t *current_brk = 0;
uint8_t *initial_brk;
uint8_t *new_brk;
uint8_t *old_brk;
int failure = 0;
int i;
void test_brk(int increment, int expected_result) {
new_brk = (uint8_t *)syscall(SYS_brk, current_brk + increment);
if ((new_brk == current_brk) == expected_result)
failure = 1;
current_brk = (uint8_t *)syscall(SYS_brk, 0);
}
void test_result() {
if (!failure)
puts("OK");
else {
puts("failure");
exit_status = EXIT_FAILURE;
}
}
void test_title(const char *title) {
failure = 0;
printf("%-45s : ", title);
fflush(stdout);
}
test_title("Initialization");
test_brk(0, 1);
initial_brk = current_brk;
test_result();
test_title("Don't overlap \"brk\" pages");
test_brk(HOST_PAGE_SIZE, 1);
test_brk(HOST_PAGE_SIZE, 1);
test_result();
/* Preparation for the test "Re-allocated heap is initialized". */
old_brk = current_brk - HOST_PAGE_SIZE;
memset(old_brk, 0xFF, HOST_PAGE_SIZE);
test_title("Don't allocate the same \"brk\" page twice");
test_brk(-HOST_PAGE_SIZE, 1);
test_brk(HOST_PAGE_SIZE, 1);
test_result();
test_title("Re-allocated \"brk\" pages are initialized");
for (i = 0; i < HOST_PAGE_SIZE; i++) {
if (old_brk[i] != 0) {
printf("(index = %d, value = 0x%x) ", i, old_brk[i]);
failure = 1;
break;
}
}
test_result();
test_title("Don't allocate \"brk\" pages over \"mmap\" pages");
new_brk = mmap(current_brk, HOST_PAGE_SIZE / 2, PROT_READ, MAP_PRIVATE | MAP_ANONYMOUS | MAP_FIXED, -1, 0);
if (new_brk == (void *) -1)
puts("unknown");
else {
test_brk(HOST_PAGE_SIZE, 0);
test_result();
}
test_title("All \"brk\" pages are writable (please wait)");
if (munmap(current_brk, HOST_PAGE_SIZE / 2) != 0)
puts("unknown");
else {
while (current_brk - initial_brk < 2*1024*1024*1024UL) {
old_brk = current_brk;
test_brk(HOST_PAGE_SIZE, -1);
if (old_brk == current_brk)
break;
for (i = 0; i < HOST_PAGE_SIZE; i++)
old_brk[i] = 0xAA;
}
puts("OK");
}
test_title("Maximum size of the heap > 16MB");
failure = (current_brk - initial_brk) < 16*1024*1024;
test_result();
exit(exit_status);
}
Changes introduced in patch v2:
* extend the "brk" test-suite embedded within the commit message;
* heap contents have to be initialized to zero, this bug was
exposed by "tst-calloc.c" from the GNU C library;
* don't [try to] allocate a new host page if the new "brk" is
equal to the latest allocated host page ("brk_page"); and
* print some debug information when DEBUGF_BRK is defined.
Signed-off-by: Cédric VINCENT <cedric.vincent@st.com>
Reviewed-by: Christophe Guillon <christophe.guillon@st.com>
Cc: Riku Voipio <riku.voipio@iki.fi>
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
Since mmap() with MAP_FIXED will map over the top of existing mappings,
it's a bad idea to use it to implement brk(), because brk() with a
large size is likely to overwrite important things like qemu itself
or the host libc. So we drop MAP_FIXED and handle "mapped but at
different address" as an error case instead.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
This patch adds support for running s390x binaries in the linux-user emulation
code.
Signed-off-by: Ulrich Hecht <uli@suse.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
Function bzero is deprecated, so replace it by function memset.
Signed-off-by: Stefan Weil <weil@mail.berlios.de>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
The kernel doesn't fill the buffer provided to sched_getaffinity
with zero bytes, so neither should QEMU.
Signed-off-by: Mike McCormack <mj.mccormack@samsung.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
Zeroing of the cpu array should start from &cpus[kernel_ret]
not &cpus[num_zeros_to_fill].
This fixes a crash in EFL's edje_cc running under qemu-arm.
Signed-off-by: Mike McCormack <mj.mccormack@samsung.com>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Acked-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
Quite a number of uid/gid related syscalls are only defined on systems
with USE_UID16 defined. This is apperently based on the idea that these
system calls would never be called on non-UID16 systems. Make these
syscalls available for all architectures that define them.
drop alpha hack to support selected UID16 syscalls. MIPS and PowerPC
were also defined as UID16, to get uid/gid syscalls available, drop
this error as well.
Change QEMU to reflect this.
Cc: Ulrich Hecht <uli@suse.de>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Alexander Graf <agraf@suse.de>
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
We keep a list of host architectures that do llseek with the same
syscall as lseek. S390x is one of them, so let's add it to the list.
Original-patch-by: Ulrich Hecht <uli@suse.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
The result needs to be converted as it is stored in an array of struct
ifreq and sizeof(struct ifreq) differs according to target and host
alignment rules.
This patch allows to execute correctly the following program on arm
and m68k:
#include <stdio.h>
#include <sys/ioctl.h>
#include <net/if.h>
#include <alloca.h>
#include <string.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <arpa/inet.h>
int main(void)
{
int s, ret;
struct ifconf ifc;
int i;
memset( &ifc, 0, sizeof( struct ifconf ) );
ifc.ifc_len = 8 * sizeof(struct ifreq);
ifc.ifc_buf = alloca(ifc.ifc_len);
s = socket( AF_INET, SOCK_DGRAM, 0 );
if (s < 0) {
perror("Cannot open socket");
return 1;
}
ret = ioctl( s, SIOCGIFCONF, &ifc );
if (s < 0) {
perror("ioctl() failed");
return 1;
}
for (i = 0; i < ifc.ifc_len / sizeof(struct ifreq) ; i ++) {
struct sockaddr_in *s;
s = (struct sockaddr_in*)&ifc.ifc_req[i].ifr_addr;
printf("%s\n", ifc.ifc_req[i].ifr_name);
printf("%s\n", inet_ntoa(s->sin_addr));
}
}
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
PTHREAD_STACK_MIN (16KB) is somewhat inadequate for a new stack for new
QEMU threads. Set new limit to 256K which should be enough, yet doesn't
increase memory pressure significantly.
Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
Reviewed-by: Nathan Froyd <froydnj@codesourcery.com>
Correct the broken attempt to calculate the third argument
to unlock_user() in the code path which unlocked the pollfd
array on return from poll() and ppoll() emulation. (This
only caused a problem if unlock_user() wasn't a no-op, eg
if DEBUG_REMAP is defined.)
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
When emulating a 32 bit Linux user-mode program on a 64 bit target
we implement the llseek syscall in terms of lseek. Correct a bug
which meant we were silently casting the result of host lseek()
to a 32 bit integer as it passed through get_errno() and thus
throwing away the top half.
We also don't try to store the result back to userspace unless
the seek succeeded; this matches the kernel behaviour.
Thanks to Eoghan Sherry for identifying the problem and suggesting
a solution.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Support the epoll family of syscalls: epoll_create(), epoll_create1(),
epoll_ctl(), epoll_wait() and epoll_pwait(). Note that epoll_create1()
and epoll_pwait() are later additions, so we have to test separately
in configure for their presence.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
Some architectures (like Blackfin) only implement ppoll (and skip poll).
So add support for it using existing poll code.
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
Add a configure check for the existence of linux/fiemap.h and the
IOC_FS_FIEMAP ioctl. This fixes a compilation failure on Linux
systems which don't have that header file.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Implement the FS_IOC_FIEMAP ioctl using the new support for
custom handling of ioctls; this is needed because the struct
that is passed includes a variable-length array.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
Some ioctls (for example FS_IOC_FIEMAP) use structures whose size is
not constant. The generic argument conversion code in do_ioctl()
cannot handle this, so add support for implementing a special-case
handler for a particular ioctl which does the conversion itself.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
Implement the missing syscalls sync_file_range and sync_file_range2.
The latter in particular is used by newer versions of apt on Ubuntu
for ARM.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
n setsockopt, the socket level options are translated to the hosts'
architecture before the real syscall is called, e.g.
TARGET_SO_TYPE -> SO_TYPE. This patch does the same with getsockopt.
Tested on a x86 host emulating MIPS. Without it:-
$ grep getsockopt host.strace
31311 getsockopt(3, SOL_SOCKET, 0x1007 /* SO_??? */, 0xbff17208,
0xbff17204) = -1 ENOPROTOOPT (Protocol not available)
With:-
$ grep getsockopt host.strace
25706 getsockopt(3, SOL_SOCKET, SO_ERROR, [0], [4]) = 0
Whitespace cleanup: Riku Voipio
Signed-off-by: Jamie Lentin <jm@lentin.co.uk>
Signed-off-by: Riku Voipio <riku.voipio@iki.fi>
Running programs that create large numbers of threads, such as this
snippet from libstdc++'s pthread7-rope.cc:
const int max_thread_count = 4;
const int max_loop_count = 10000;
...
for (int j = 0; j < max_loop_count; j++)
{
...
for (int i = 0; i < max_thread_count; i++)
pthread_create (&tid[i], NULL, thread_main, 0);
for (int i = 0; i < max_thread_count; i++)
pthread_join (tid[i], NULL);
}
in user-mode emulation will quickly run out of memory. This is caused
by a failure to free memory in do_syscall prior to thread exit:
/* TODO: Free CPU state. */
pthread_exit(NULL);
The first step in fixing this is to make all TaskStates used by QEMU
dynamically allocated. The TaskState used by the initial thread was
not, as it was allocated on main's stack. So fix that, free the
cpu_env, free the TaskState, and we're home free, right?
Not exactly. When we create a thread, we do:
ts = qemu_mallocz(sizeof(TaskState) + NEW_STACK_SIZE);
...
new_stack = ts->stack;
...
ret = pthread_attr_setstack(&attr, new_stack, NEW_STACK_SIZE);
If we blindly free the TaskState, then, we yank the current (host)
thread's stack out from underneath it while it still has things to do,
like calling pthread_exit. That causes problems, as you might expect.
The solution adopted here is to let the C library allocate the thread's
stack (so the C library can properly clean it up at pthread_exit) and
provide a hint that we want NEW_STACK_SIZE bytes of stack.
With those two changes, we're done, right? Well, almost. You see,
we're creating all these host threads and their parent threads never
bother to check that their children are finished. There's no good place
for the parent threads to do so. Therefore, we need to create the
threads in a detached state so the parent thread doesn't have to call
pthread_join on the child to release the child's resources; the child
does so automatically.
With those three major changes, we can comfortably run programs like the
above without exhausting memory. We do need to delete 'stack' from the
TaskState structure.
Signed-off-by: Nathan Froyd <froydnj@codesourcery.com>
Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
On many systems, socklen_t is defined as unsigned. This means that
checks for negative values are not meaningful.
Fix by explicitly casting to a signed integer.
This also avoids some warnings with GCC flag -Wtype-limits.
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
When loading a shared library that requires an executable stack,
glibc uses the mprotext PROT_GROWSDOWN flag to achieve this.
We don't support PROT_GROWSDOWN.
Add a special case to handle changing the stack permissions in this way.
Signed-off-by: Paul Brook <paul@codesourcery.com>
There's no _llseek on s390x either. Replace the existing
test for __x86_64__ with a functional test for __NR_llseek.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Libc will fallback gracefully if pselect6 is not available. Thus put
pselect6 to nowarn until the atomicity issues of the original pselect6
patch are dealt with.
Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
Cc: Michael Casadevall <mcasadevall@ubuntu.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Alpha passes oldset by value in a register, and returns the newset
as the return value; as compared to the standard implementation in
which both are passed by reference. This requires being able to
distinguish negative return values that are not errors. Do this in
the same way as the Alpha Linux kernel, by storing a zero in V0 in
the implementation of the syscall.
At the same time, fix a think-o in the regular sigprocmask path in
which we passed the target, rather than the host, HOW value.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Alpha passes the signal set in a register, not by reference.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
At the same time, tidy the code wrt MIPS and SH4 which have the
same two register return mechanism. Fix confusion between pipe
and pipe2 with an explicit flags=0, when the guest will not be
using the two register return mechanism.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
rlim_t conversion between host and target added.
Otherwise there are some incorrect case like
- RLIM_INFINITY on 32bit target -> 64bit host.
- RLIM_INFINITY on 64bit host -> mips and sparc target ?
- Big value(for 32bit target) on 64bit host -> 32bit target.
One is added into getrlimit, setrlimit, and ugetrlimit. It converts both
RLIM_INFINITY and value bigger than target can hold(>31bit) to RLIM_INFINITY.
Another one is added to guest_stack_size calculation introduced by
703e0e89. The rule is mostly same except the result on the case is keeping
the value of guest_stack_size.
Slightly tested for SH4, and x86_64 -linux-user on x86_64-pc-linux host.
Signed-off-by: Takashi YOSHII <takasi-y@ops.dti.ne.jp>
Acked-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Check TARGET_ABI_BITS, not TARGET_LONG_BITS, when deciding
whether or not the guest needs special 64-bit stat translation.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
2nd arg of page_set_flags() should be start+size, but size.
Signed-off-by: Takashi YOSHII <takasi-y@ops.dti.ne.jp>
Acked-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Commit c05c7a7306
breaks cross compilation for mips (and other
compilations without CONFIG_INOTIFY1):
make[1]: Entering directory `/qemu/bin/mips'
CC i386-linux-user/syscall.o
cc1: warnings being treated as errors
/qemu/linux-user/syscall.c: In function ‘do_syscall’:
/qemu/linux-user/syscall.c:7067: error: implicit declaration of function ‘sys_inotify_init1’
Cc: Riku Voipio <riku.voipio@nokia.com>
Cc: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Stefan Weil <weil@mail.berlios.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
ia64 has some strangenesses that need to be workaround:
- it has a __clone2() syscall instead of the using clone() one, with
different arguments, and which is not declared in the usual headers.
- ucontext.uc_sigmask is declared with type long int, while it is
actually of type sigset_t.
- uc_mcontext, uc_sigmask, uc_stack, uc_link are declared using #define,
which clashes with the target_ucontext fields. Change their names to
tuc_*, as already done for some target architectures.
New syscall which gets actively used when you have a
fresh kernel.
Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
On linux/sh4
pipe() return values by r0:r1 as SH C calling convention.
pipe2() return values on memory as traditional unix way.
Signed-off-by: Takashi YOSHII <takasi-y@ops.dti.ne.jp>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Move userland PALcode handling into linux-user main loop so that
we can send signals from there. This also makes alpha_palcode.c
system-level only, so don't build it for userland. Add defines
for GENTRAP PALcall mapping to signals.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
