GraphicalUserInterface:
- The Show() method now checks if the team window is already visible. If
it is, we instead call Activate() to focus it. Fixes the problem that
attempting to attach to a team which we already had a running team
debugger for would appear to do nothing, rather than focusing that window.
TeamDebugger:
- Add accessor to query whether the debugger's interface is a post mortem core
rather than a live team.
TargetHostInterface:
- When asked to locate a team debugger by team_id, as is done for attach
requests to determine if we already have an existing instance attached, skip
over post mortems. This takes care of the potential problem of detecting a
core instance with the same team_id as a live team, and consequently refusing
the attach request.
- When a team debugger quits, use its actual pointer to look it up rather than its
team ID, as we could otherwise potentially remove the wrong one in the case of a
clash.
{Local}TargetHostInterface:
- Add virtual method for requesting a debugger interface for a core file.
Implement accordingly in LocalTargetHostInterface based on Ingo's
previous implementation in the Debugger app.
TeamDebuggerOptions:
- Add an enum to define the type of team debugger request being made
to explicitly make this clear rather than guessing based on the provided
arguments. Set accordingly in the various cases.
- Add a parameter for the core file path.
Debugger:
- Refactor to fill in TeamDebuggerOptions appropriately for core file
requests, and consequently simplify code.
This doesn't yet deal with the fact that a post-mortem team's ID can
clash with that of a live team, which may lead to issues when attempting
to attach to a live team in such a case.
* The latter doesn't link against libroot.so, which lets the library init fail.
* Fix kernel_debugger crash in malloc().
* libsupc++.so will be deleted from gcc_syslibs package (the static libs will
stay however).
* DebuggerInterface: Add method IsPostMortem() to be able to
descriminate between live and post mortem debugging.
* Add DebuggerInterface implementation CoreFileDebuggerInterface which
provides information from a core file.
* TeamDebugger: Don't start the debug event thread when debugging post
mortem.
* Debugger: New command line variant "Debugger --core <file path>",
which starts a team debugger using the core file.
There are a few issues:
* I didn't see an easy way to integrate with the new
TargetHostInterface framework and I didn't want to get into Rene's
way changing stuff. As a side effect core file debug windows are not
counted and Debugger will quit when only those are left, respectively
will additionally open a teams window on start-up.
* There aren't any symbols yet. We can't use the debug kit
functionality, since it isn't bitness/endianess agnostic. So either
it needs to be adjusted or ported over to Debugger.
* It was previously geared towards shared object files. Now it is
fairly agostic and also accepts e.g. files without sections and
different endianess.
* The handling of 32 vs. 64 bit files works a bit differently now.
There're structs ElfClass32 and ElfClass64 now which provide the
types for the respective bitness.
* Add a few more getters.
* Add function core_dump_write_core_file(). It writes a core file for
the current thread's team. The file format is similar to that of
other OSs (i.e. ELF with PT_LOAD segments and a PT_NOTE segment), but
most of the notes are Haiku specific (infos for team, areas, images,
threads). More data will probably need to be added.
* Add team flag TEAM_FLAG_DUMP_CORE, thread flag
THREAD_FLAGS_TRAP_FOR_CORE_DUMP, and Team property coreDumpCondition,
a condition variable available while a core dump is progress. A
thread that finds its flag THREAD_FLAGS_TRAP_FOR_CORE_DUMP set before
exiting the kernel to userland calls core_dump_trap_thread(), which
blocks on the condition variable until the core dump has finished. We
need the team's threads to stop so we can get their CPU state (and
have a generally unchanging team state while writing the core file).
* Add user debugger message B_DEBUG_WRITE_CORE_FILE. It causes
core_dump_write_core_file() to be called for the team.
* Dumping core as an immediate effect of a terminal signal has not been
implemented yet, but that should be fairly straight forward.
Similar to arch_get_debug_cpu_state(), but the thread whose CPU state
to retrieve is specified. Works only for threads that aren't running,
and on x86-64 we can get the FPU state only when the thread was
interrupted in userland.
Not implemented for the incomplete architecture ports.
This resolves a TODO: We used thread_interrupt() to wake up the thread
from an interruptable wait. However, if the thread was already in the
kernel and about to start waiting, that would have no effect and the
thread would wait anyway. Now there's the new non-blockable signal
SIGNAL_DEBUG_THREAD, which is sent to the thread instead, making sure
that thread doesn't start waiting.
* New Intel SkyLake seems to have 9 mapped ranges
at boot. It seems like this define has been creeping
up for a while.
* Resolves the inital issue reported in #11377 on SkyLake
as well. Bonefish mentioned it might need to be raised
again... he had some good foresight there :-)
* I'm seeing the same no bootable partitions issue though
via USB after this raise. (maybe a USB 3.1 thing?)
* Previous enablement broke gcc2 and hybrid builds
* gcc2 builds fixed and tested working now
* Fix Hybrid builds via missing MultiArch Jam
* Sorry for the noise, enjoy early Bluetooth support
* Makes the shutdown process to rely on the instance that was
running at the time the function is called. While it's unlikely
anyone will see any change from that, this way we will not conflict
in any case with the launch_daemon.
* The reply could have been confused in a remote situation, make
every request to use an it's own object.
* In launch_media_server there was a wrong variable redefinition
too.
* Also fix a possible issue with List::Get, the code returned
after a Get() error. While I doubt if this happens the next
items will be still valid, I prefer to continue until we reach
the end of the items, and in case of a SyncToRoster request we will
at least get the list empty. We might have a blocking call
on the other end, so it's our own duty to at least try to avoid
this in any case. As final result at least we will get the list
empty.
* When the user isn't requesting a custom notification, it will
be a BMediaRoster job to do it.
* Reintroduce BMediaRoster::SyncToServices, this time based on local
message passing rather than a global semaphore.
* SyncToServices is used in launch_media_server to make the process
more launch_daemon safe and faster in the average case.
* It was an error to add notifications in the media_server.
* Fixes#12717.
TargetHostInterfaceRoster:
- Add Listener interface. For now, this simply notifies the listener of
changes to the active debugger count.
- Adjust show new team window command to automatically fall back to the local
interface window if one isn't specified. Fixes the Start New Team menu item
in the TeamWindow. The latter will later be expanded to show the available
interfaces to start a new team on in a submenu.
Debugger:
- Implement roster listener interface in order to know when to attempt
application quit.
With this commit, all necessary work to isolate the application from the target
host is complete, and work on the actual remote interface and protocol can
begin.
TargetHostInterfaceRoster:
- Implement TargetHostInterface's Listener interface. This allows us
to more cheaply track the total number of running team debuggers,
as well as automatically removing an interface that is destroyed.
Application objects:
- Rework and simplify to take into account that they will no longer be
directly managing the team debugger list. Requests to start a new debugger
are still funnelled through here however, and as such, said requests must
now provide the appropriate target host to start with. Adjust StartTeamWindow
and TeamsWindow accordingly.
- On global init, always create an instance of the local interface.
TargetHostInterface:
- Convert to BLooper and implement TeamDebugger's Listener interface.
TargetHostInterfaces now directly manage their TeamDebugger instances,
and consequently take over the equivalent duties that the main application
previously had.
- Adjust signatures of Attach/CreateTeam to add const. Adjust
LocalTargetHostInterface accordingly.
- Add accessor to determine if a given interface is local or not. Will be
needed for the TeamDebugger's file manager eventually so it knows if it
needs to request remote files if no matching local file is found.
- Add accessor to start a TeamDebugger instance, and corresponding options
structure.
TargetHostInterfaceRoster:
- Minor adjustments to host interface initialization to take into account
needing to start the looper.
- Add accessor for number of running team debuggers, for the main app to
use when deciding to quit.
TeamDebugger:
- Add accessor for SettingsManager. Needed for the case of a restart request,
as the host interfaces do not have direct access to it.
TeamsWindow:
- For now, always grab the local host interface when initializing the window.
Once the remote interface is implemented, this will need to be adjusted, but
the appropriate UI for creating/selecting it is needed first anyways.
With these changes, the main application is fully host-agnostic, and all
management of actual debuggers is delegated to their parent host interfaces.
There still needs to be a listener interface for the host interface and/or
roster though, so that the application can be made aware of when debuggers
quit, as this drives whether it's time to terminate the app or not.
* Intel panel scaling was making native mode blury
* Resolutions < native result in a non-scaled screen for now.
* We should look into using the hardware scaler vs
doing fake scaling.
* Resolves#12716
LocalTargetHostInterface:
- Implement the create and attach functionality. In theory, this completes
everything that's needed in order to adjust the main application to do
all debugger interface creation via the roster.
Also make some new const variables to make some math more opaque
i.e. don't use as many magic constants.
This works great at 12pt font size, decenly at all other sizes. There is a
bit of jitter at 13pt and 24pt for some reason when moving from a default
settings view to another settings view.
Ingo Weinhold