When iterator->current is NULL, hash_next() assumes we've reached the
end of a bucket (linked list) and moves to the next one. Wehn the first
element of a linked list was removed in hash_remove_current()
iterator->current was set to NULL, causing the next call to hash_next()
to skip over the rest of the list of that bucket.
To fix this we now decrement iterator->bucket by one, so the next call
to hash_next() correctly arrives at the new first element of the same
bucket. Doing it this way avoids having to search backwards through the
table to find the actual previous item.
This caused modules to be skipped in module_init_post_boot_device()
when normalizing module image paths so some of the module images ended
up non-normalized. This could then cause images to be loaded a second
time for modules that were part of an actually already loaded image.
This setup is present for the PCI module with the pci/x86 submodule
and would lead to a second copy of the PCI module image to be loaded
but without being initialized, eventually leading to #8684.
The affected module images were pretty much random, as it depended on
the order in which they were loaded from the file system, in this case
the boot floppy archive of the El-Torito boot part of ISO and anyboot
images. The r1alpha4 release images unfortunately had the module files
ordered in the archive just so that the PCI module image would be
skipped, allowing #8684 to happen on many systems with MSI support.
Since the block cache uses hash_remove_current() as well in some cases,
it is possible that transactions in its list could've been skipped.
Cursory testing didn't reveal this to be a usual case, and it is
possible that in the pattern it is used there, the bug wouldn't be
triggered at all. It's still possible that it caused rare misbehaviour
though.
- BreakpointTableModel now encapsulates both breakpoints and watchpoints.
- Extended BreakpointView and TeamWindow to handle enable/disable/removing
watchpoints as well.
This comes from a similar patch in FreeBSD r234666 but does the check
and acknowledge in the case where we don't have any other interrupts.
Since the interrupt isn't used it is never unmasked and therefore the
check against the interrupt mask will always fail if the TWSI IRQ is
the only interrupt that happened. In that case an interrupt storm would
be triggered that lasts until any other valid interrupt comes along, is
handled and both are acknowledged.
This may help with #8454 on some cards that exhibit the spurious TWSI
IRQs.
Since we now implement HAIKU_CHECK_DISABLE_INTERRUPTS there is no need
to do it again in the interrupt handler. Further, store and carry over
the interrupt status from the hook to the interrupt handler as that is
cleaner and just in case reading the register somehow clears it. May
help with #8454.
Move calculating the width of the column title itself out to
OutlineView::GetColumnPreferredWidth(). Previously, each pass would
compute the width of both the field itself and the column title,
leading to considerable redundant work. Also, take outline level indent
into account in the resulting width. Should improve performance a bit.
* Before, you had to have both, the text view layout item, and the label
layout item or else nothing would ever be visible.
* Now you can only create the text view item, and it will still work.
* Also, no matter the order you added the layout items, they would always
put the label on the left, and the control to the right.
* You can place the label and text view layout items anywhere now, although
you should keep in mind that the view spans over their frame unions; IOW
they should always adjacent to each other, but not necessarily horizontally
and left to right.
* No longer uses a fixed label spacing, but utilizes
BControlLook::DefaultLabelSpacing() instead.
* However, the spacing is always added to the right of the label, no matter
how you place it in the layout. Maybe one wants to add a SetLabelTextViewGap()
like method.
* Adjust BTextView to use B_COMMAND_KEY instead of B_CONTROL_KEY
for wordwise navigation and jumping to the top and bottom.
This requires a shortcut, which is only installed if there is
none already (for the groups B_LEFT_ARROW/B_RIGHT_ARROW and
B_HOME/B_END). As a result, wordwise navigation no longer works
in Mail, for instance.
* drop "protected" from bsd-compat header sys/cdefs.h, as that define
pollutes the global namespace and at least FreeBSD doesn't provide
it anymore
* remove all uses of that macro from libedit, which seems to be the
only user in our tree
* When in full screen mode, add a tooltip with the window title,
and a keyboard shortcut hint so one knows how to get out of it.
Should help with #7356
* 80 column cleanup
* Fixed issue with unwanted keymap switching in case UnZip started in
* background (expanding optional packages during Haiku build, for
* example). UnZip executable has no background application flag
* for unknown reason.
+alpha4
This gets basic watchpoint support working. Right clicking on a variable
and picking Watch now opens a prompt with the inferred address, size
and watch type for the user to adjust.
Still needs some work to get them to show/be modifiable in the breakpoints
tab and to get them to respect architectural restrictions (i.e. on x86 we
can realistically only do 2 hardware watchpoints and those are restricted
to write watch), at least until we support software emulated watchpoints.
