If QEMU forks after the CPU threads have been created, qemu_mutex_lock_iothread
will not be able to do qemu_cpu_kick_thread. There is no solution other than
assuming that forks after the CPU threads have been created will end up in an
exec. Forks before the CPU threads have been created (such as -daemonize)
have to call rcu_after_fork manually.
Notably, the oxygen theme for GTK+ forks and shows a "No such process" error
without this patch.
This patch can be reverted once the iothread loses the "kick the TCG thread"
magic.
User-mode emulation does not use the iothread, so it can also call
rcu_after_fork.
Reported by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Tested by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
New threads always point at the same env which is incorrect and usually
leads to a crash.
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
Reviewed-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
The second and fourth argument are in/out parameters, store them back
after the syscall. Also, the fourth argument was mishandled, and EFAULT
handling was missing.
Signed-off-by: Andreas Schwab <schwab@suse.de>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
start/end_exclusive() need be pairs, except the start_exclusive() in
stop_all_tasks() which is only used by force_sig(), which will be abort.
So at present, start_exclusive() in stop_all_task() need not be paired.
queue_signal() may call force_sig(), or return after kill pid (or queue
signal). If could return from queue_signal(), stop_all_task() would not
be called in time, the next end_exclusive() would be issue.
So in arm_kernel_cmpxchg64_helper() for ARM, need remove end_exclusive()
after queue_signal(). The related commit: "97cc756 linux-user: Implement
new ARM 64 bit cmpxchg kernel helper".
Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
When support was added for TrustZone to ARM CPU emulation, we failed
to correctly update the support for the linux-user implementation of
the get/set_tls syscalls. This meant that accesses to the TPIDRURO
register via the syscalls were always using the non-secure copy of
the register even if native MRC/MCR accesses were using the secure
register. This inconsistency caused most binaries to segfault on startup
if the CPU type was explicitly set to one of the TZ-enabled ones like
cortex-a15. (The default "any" CPU doesn't have TZ enabled and so is
not affected.)
Use access_secure_reg() to determine whether we should be using
the secure or the nonsecure copy of TPIDRURO when emulating these
syscalls.
Signed-off-by: Mikhail Ilyin <m.ilin@samsung.com>
Message-id: 1426505198-2411-1-git-send-email-m.ilin@samsung.com
[PMM: rewrote commit message to more clearly explain the issue
and its consequences.]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This reverts commit b8a173b25c, reversing
changes made to 5de090464f.
(I applied this pull request when I should not have done so, and
am now immediately reverting it.)
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This was the only caller of cpu_init() that was not checking for NULL
yet.
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Simple "hello world" MIPS N32 userland program crashes with segfault due to
incorrectly defined stat structure in QEMU.
Correct "target_stat" definition to match kernel's "stat64" as in MIPS N32
there are only plain "stat" syscalls using 64-bit structure.
Reported-by: Daniel Sanders <daniel.sanders@imgtec.com>
Signed-off-by: Leon Alrae <leon.alrae@imgtec.com>
Tested-by: Daniel Sanders <daniel.sanders@imgtec.com>
Reviewed-by: James Hogan <james.hogan@imgtec.com>
Fix TARGET_SI_PAD_SIZE calculation to match the way the kernel does it.
Use different TARGET_SI_PREAMBLE_SIZE for 32-bit and 64-bit targets.
Signed-off-by: Maxim Ostapenko <m.ostapenko@partner.samsung.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In abi_long do_ioctl_dm(), after lock_user() call, the code does
not call unlock_user() before going to failure return in default case.
Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In main.c, all SIG* should be TARGET_SIG*, since the relevant functions
(queue_signal() and gdb_handlesig()) expect TARGET_SIG*.
The corresponding vi command is "1,$ s/\<SIG/TARGET_SIG/g".
Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When failure occurs during locking of vec[i], we also need to unlock all
already locked vec[i] in failure processing code block before return.
Code in unlock_user() checks vec[i].iov_base for NULL, so there's no
need not check it .
If error is EFAULT when "i == 0", vec[i].iov_base is NULL, we can just
skip it, so can still use "while (--i >= 0)" loop condition.
Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
When passing ancillary data through a unix socket, handle
credentials properly instead of doing a simple copy and
issuing a warning.
Signed-off-by: Alex Suykov <alex.suykov@gmail.com>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
The m68k signal frame setup code which writes the signal return
trampoline code to the stack was assuming that a 'long' was 32 bits;
on 64 bit systems this meant we would end up writing the 32 bit
(2 insn) trampoline sequence to retaddr+4,retaddr+6 instead of
the intended retaddr+0,retaddr+2, resulting in a guest crash when
it tried to execute the invalid zero-bytes at retaddr+0.
Fix by using uint32_t instead; also use uint16_t rather than short
for consistency. This fixes bug LP:1404690.
Reported-by: Michel Boaventura
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
Without this fix, qemu segfaults when emulating the sigaltstack syscall,
because it incorrectly treats the ss_flags field as 64 bits rather than 32
bits.
Signed-off-by: Ed Swierk <eswierk@skyportsystems.com>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
linux-user passes the cmd argument of the ioctl syscall as a signed long,
but compares it to an unsigned int when iterating through the ioctl_entries
list. When the cmd is a large value like 0x80047476 (TARGET_TIOCSWINSZ on
mips64) it gets sign-extended to 0xffffffff80047476, causing the comparison
to fail and resulting in lots of spurious "Unsupported ioctl" errors.
Changing the target_cmd field in the ioctl_entries list to a signed int
causes those values to be sign-extended as well during the comparison.
Signed-off-by: Ed Swierk <eswierk@skyportsystems.com>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
The resource argument is translated from host to target for
[gs]etprlimit but not for prlimit64. Fix this.
Signed-off-by: Felix Janda <felix.janda@posteo.de>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
The function copy_siginfo_to_user() just calls tswap_siginfo(), so
call the latter function directly and delete the wrapper function.
The wrapper is actually misleading since it implies that the
semantics are like the kernel function with the same name which
copies the data to a guest user-space address. In fact tswap_siginfo()
just does data-structure conversion between two structures whose
addresses are host addresses (the copy to userspace is handled
in QEMU by the lock_user/unlock_user calls).
This also fixes clang complaints about the wrapper being unused
in some configs.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
The function end_exclusive() isn't used on all targets; mark it as
such to avoid a clang warning.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
