This hack was introduced in issue#287 which later becomes endless maintainance pain.
=====
Our previous check_exit_request use `brcond` in the middle of a TranslationBlock which
breaks the assumptions and thus a hack to liveness_pass_1 is used for _all_ brcond instructions
which causes issues for MIPS and many other scenarios.
=====
This patch also resolves PC not sync-ed when no memory hooks are installed, finally. Now
Unicorn will always have correct PC no matter what happens.
* use hwaddr for paddrs
* Fix the truncation for memory hooks as well
* Add LPAE regression test
Co-authored-by: Takacs, Philipp <philipp.takacs@iosb.fraunhofer.de>
* memory snapshots fix use after free on flatview copy
When restoring a snapshot with memory the flatview must be restored
before the memory reagions are filtered. Because the
AddressSpaceDispatcher also has pointer to the MemoryRegions and on copy
they need to be cleared. The memory_filter_subregions function frees
MemoryRegions which are not used at the time of the snapshot.
* fix some memleaks in tests
These tests has forgott to call uc_close(uc), which lead to memory
leaks. Found by the LeakSanitizer.
* memory snapshots correct clean up container memory regions
* Fix further stackoverflow in tests
---------
Co-authored-by: mio <mio@lazym.io>
* Support mips64 - write correct pc register width on uc_emu_start
* Convert to UC_MODE_MIPS64
* Correctly select MIPS64 CPU model
* Simple 64-bit test - check it doesn't crash
* lint
* Comment
* Comment
* Add offset when indexing cpu model, makes tests work on older python
* Move test
* add PC check to test
* Fix test - add python version check
* Use RegressTest method for assert
Rationale: Previouly, Unicorn uses several hacks to pretend it supports
floating point instructions while not properly setting up something
like CPU features. Therefore, once related registers like CR4 is reset,
the hacks stop working and UC_ERR_INSN_INVALID is thrown. Setting the default
model to a CPu that has basical floating point support should have the
minimal break changes.
This code was commented out since 2021, but by default, the error
codewas initialized to `UC_REG_OK`, so there was no error returned
untila result, any write to `UC_ARM_REG_C1_C0_2` returned an error.
for glibc, if _GNU_SOURCE is defined, <time.h> will include <sys/timex.h>
but not for musl, so add "#include <sys/timex.h>"
although `man clock_adjtime` said "#include <sys/timex.h>", but it won't
work for glibc w/o "#include <time.h>", I don't known why yet.
PS it seems clock_adjtime is used nowhere?
Signed-off-by: Z. Liu <zhixu.liu@gmail.com>
* qemu/tcg: fix UC_HOOK_MEM_READ on aarch64.
Directly jump into the slow path when there is any hookmem enabled. This
fixes#1908.
Signed-off-by: Glenn Baker <glenn.baker@gmx.com>
* qemu/tcg: fix UC_HOOK_MEM_READ on ppc64.
Directly jump into the slow path when there is any hookmem enabled.
Signed-off-by: Glenn Baker <glenn.baker@gmx.com>
* qemu/tcg: check for UC_HOOK_MEM_READ_AFTER.
Use has_hookmem() helper to determine wether "slow-path" TLB read is
needed. Add this helper to x86 architecture as well so that to check for
all hookmem.
Signed-off-by: Glenn Baker <glenn.baker@gmx.com>
* qemu/tcg: factor out has_hookmem().
It's the same implementation for all architectures, so factor out
has_hookmem() into tcg_uc_has_hookmem().
Signed-off-by: Glenn Baker <glenn.baker@gmx.com>
---------
Signed-off-by: Glenn Baker <glenn.baker@gmx.com>
