On 64 bit platforms a 64 bit size_t was written at the (incorrect)
uint32 on the stack, causing the adjacent bytes variable to be
clobbered. Because of this, the vectors wouldn't actually be filled
with any file data, making the content of files inacessible.
It seems getting the device from the partition after it is unmounted
does not work reliably. Get the device much earlier in the function,
before actually unmounting it.
Fixes#2630.
* Incorrect handling of realloc() failure (if reallocation fails,
original buffer is still allocated and must be freed),
* Use of bit shift on signed integer (undefined behavior in C and C++).
Calling Debugger here causes issues with some software (mainly Mesa).
Only print a warning, this is enough for developers to see the problem,
and not as annoying for users.
The Be Book [1] states that these are supposed to be C functions,
but the comment (see diff) said they were deprecated, and the fact
that they were in a .cpp file (and I checked with "nm") made them
C++ functions.
I can't recall any applications failing to start because these
symbols were unresolved, so we might be able to just remove them...
[1]: https://www.haiku-os.org/legacy-docs/bebook/BFilePanel_Overview.html
"git blame" indicates they've been here since Tracker was imported
into the tree, and a single paragraph explaining their existence
just states they are more efficient/faster than the ones in BString
due to running simple checks before calling str*cmp.
If there is (still) such a performance problem in our BString, we should
fix it there and not use hacky workarounds. I didn't notice any difference
in speed when using Tracker with this patch; however, I'm on a reasonably
fast machine.
- Factor out RegistersView's SIMD output formatting functions into
UiUtils helpers. Adjust RegistersView accordingly.
- Adjust DebugReportGenerator to detect SIMD registers and format
them appropriately for report output using the aforementioned
helpers.
RegisterTableModel:
- Add member for preferred SIMD unit format, and use it to decide
what to divide up such registers as, rather than hardcoding 16 bit integer.
RegistersView:
- When right clicking on an SIMD register, display a format context menu
allowing the user to decide what packed unit format to interpret the register
data as, and adjust table model accordingly.
DebuggerInterface,Architecture{X86,X8664}:
- Add hook function for retrieving a feature flag mask for the target CPU,
and corresponding implementation in Architecture-specific classes.
ArchitectureX86:
- Read CPU features on init, and use them to restrict the exposed set of
registers such that the MMX and SSE registers will only be visible
if the target CPU actually supports the respective instructions.
- If an SIMD register is detected, retrieve its value and format
it as an array of 16-bit hex values. This will be extended to
allow other format choices in a future commit.
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?
