ArchitectureX86{-64}:
- Implement helper functions for adding floating point and SIMD registers.
- Add st0-st7 and mm0-mm7 to the list of exposed x86{-64} registers,
along with their DWARF unwind mappings.
- Add xmm0-xmm7 for x86, and xmm0-xmm15 for x86-64.
CpuStateX86{-64}:
- Implement helper functions for setting/retrieving floating point and SIMD
register values.
- Fill in values for st*, mm* and xmm* from debug_cpu_state.
Since zip has been outsourced, it's possible to loose it when
upgrading, in which case ZOM went busylooping without noticing the
thread actually never started.
At least now it says "failed to create the archive".
Maybe it should try to install zip?
If it was already determined that the memory is within the kernel
stack, a simple memcpy is enough.
This allows capturing kernel stack traces in situations where a fault
handler cannot be installed (i.e. where one is already installed).
When extracting/inserting report items there might not be a full uint32
available in the report anymore. Only copy as many bytes as are actually
needed by the report item and guaranteed to be present in the report.
Twice the size was copied due to a missed adjustment in hrev31839,
reading past the key state array. This didn't cause any corruption
because the overwritten state wasn't used anymore later on and the
write didn't overrun. It could cause a crash however if the read went
past the allocated area.
On close a flag is set in the cookie of this user of a protocol handler
and the device cancels its pending transfer. This wakes up any possible
listeners. When the closed flag is set, an error code is returned from
_ReadReport() which causes the retry loop to be left. Handlers listening
on the same device which were not closed just retry the transfer.
This ensures that closing a device will cause pending control requests
to complete with a sensible error code.
The concept of entry point in COFF is actually different than in ELF.
In COFF, the entry point is actually a "descriptor" (pointer) to the actual
start code. So we patch the entry point address when calling objcopy.
Now my old Performa 5400/180 actually starts the loader correctly \o/
