The LLVM Coding Standards [1] specify that:
> [T]o match error message styles commonly produced by other tools,
> start the first sentence with a lowercase letter, and finish the last
> sentence without a period, if it would end in one otherwise.
Historically, that hasn't been something we've enforced in LLDB, but in
the past year or so I've started to pay more attention to this in code
reviews. This PR brings more error messages in compliance, further
increasing consistency.
I also adopted `createStringErrorV` where it improved the code as a
drive-by for lines I was already touching.
[1] https://llvm.org/docs/CodingStandards.html#error-and-warning-messages
Assisted-by: Claude Code
The only difference between them is that OptionalBool's third state
is "unknown" and LazyBool's is "calculate". We don't need to tell
the difference in a single context, so I've made a new eLazyBoolDontKnow
which is an alias of eLazyBoolCalculate.
Avoids manual memory management.
Uses `shared_ptr` instead of `unique_ptr` because we store references to
the current thread in a backup variable.
Simplifies the private thread `is_secondary` semantics by providing a
backup storage for the current thread instead of a boolean value with a
contract to manage the backup separately.
lldb had three preprocessor defines for logging,
LLDB_LOG - formatv style argument
LLDB_LOGF - printf style argument
LLDB_LOGV - formatv style argument, only when verbose enabled
If you weren't looking at Log.h and the definition of these three, and
wanted to log something with formatv, it was easy to use LLDB_LOGV by
accident. We just had a situation where an important log statement
wasn't logging and it turned out to be this. This is fragile if you
aren't looking at the header directly, so I'd like to make this more
explicit. My proposal:
LLDB_LOG - formatv style argument
LLDB_LOG_VERBOSE - formatv style argument, only when verbose enabled
LLDB_LOGF - printf style argument
LLDB_LOGF_VERBOSE - printf style argument, only when verbose enabled
The new fouth one is to remove several places where we do `if (log &&
log->GetVerbose()) LLDB_LOGF (...)` in the sources today, and make both
styles consistent.
This PR implements that change, mechanically changing all LLDB_LOGV's to
LLDB_LOG_VERBOSE.
It also updates many of the `if (log && log->GetVerbose()) LLDB_LOGF`'s.
Some uses of this conditional expression do extra calculations in
addition to logging, and so those were left as-is so we're not doing
throwaway work when running without verbose logging.
There were many instances throughout lldb where callers are still doing
`if (log) LLDB_LOG*(...)`, a remnant of when all calls were to the `Log`
object's `Printf()` method, and you had to check if your local Log*
pointer was non-nullptr before calling the method. I removed those,
again keeping ones where work for logging is done in the block of code.
The code changes are all mechanical and uninteresting, but the question
of whether this naming change is widely agreed on is maybe worth
discussing.
This PR replaces the Get*CallbackAtIndex pattern in the PluginManager
with returning a snapshot of callbacks that the caller can iterate over
using a range-based for loop. This is a continuation of #184452 which
added thread safety by using snapshots. However, that introduced a bunch
of unnecessary copies which are largely eliminated again by getting the
snapshot once when gather all the callbacks, rather than doing that on
each iteration when querying a plugin for a given index. It also
eliminates the possibility of the snapshot changing underneath you when
iterating over the plugins.
This change was largely mechanical and I used Claude to do the menial
work of updating the signatures and call sites.
This reverts commit 97572c1860efeeb97b5940927cee72081b61810a.
This patch seems to cause TestWatchpointCommandPython.py to time out
on the ubuntu buildbots (but nowhere else that I can find so far.) The
timeout is weird too, the TEST FILE is timing out but the individual
tests aren't being shown and there's no other output. Grrr...
Anyway I'll revert this and then see if I can do some guessing about
how this change might cause the test to fail.
When lldb stops to run a breakpoint condition or other callback that has
to happen between the private stop and returning control to the user, it
will run in the state where the public state is still "running". But if
the callback needs to run lldb commands or python code, it needs to see
the correct "stop" state.
We used to handle that by switching the public state to stopped before
running the callbacks. However, that opened a window where we are still
handling the stop event and another thread would be allowed to continue
the target or do other actions that can interfere with that orderly
process.
This patch adds the ability to designate a particular thread as "seeing
the private state" while all other threads see the "public state". Then
when we run a breakpoint callback, no threads but the one that is
actually running the callback will see the state change until the event
has been delivered to the primary state listener.
It also adds a test that while a long-running breakpoint callback runs,
another thread continues to see the state as running.
In #184259, Jim noticed that Debugger::FindTargetWithProcess and
Debugger::FindTargetWithProcessID are rather poorly designed APIs as
tehy allow code running in one Debugger to mess with Targets from
another Debugger. The only use is Process::SetProcessExitStatus which
isn't actually used.
Fix asan error mentioned in [this
comment](https://github.com/llvm/llvm-project/pull/179799#issuecomment-3882710678)
This is breaking our internal builds.
Details: The field needs to be cleaned up or we'll have a leak. We could
also make it a smart pointer if you'd like. But as-is, this is the
quickest way to resolve the issue.
In #168245, I attempted to dump the available settings to Markdown. That
required a full build of LLDB. However, to build the docs, only the swig
wrappers should need to be compiled. The comment was that we should be
able to use the definitions from the TableGen files.
Currently, the property definitions in don't have information about the
path where they will be available. They only contain a `Definition`
which groups properties, so they can be added to
`OptionValueProperties`.
With this PR, I'm adding the path for each property definition. For
example, `symbols.enable-external-lookup` would have `Name =
enable-external-lookup, Path = symbols`. In LLDB itself, we don't need
this path, we only need it for the documentation. To avoid mismatches
between the actual path and the declared one, I added a debug-only check
when a property group is added to a parent
(`OptionValueProperties::AppendProperty`).
The TableGen emitter for the properties now additionally emits
`g_{definition}_properties_def`, which includes both the array of
properties and the expected path. This constant has to be used to
initialize a `OptionValueProperties`.
I couldn't test this for everything (e.g. IntelPT or ProcessKDP), but
the necessary changes are simple: (1) set the `Path` in the TableGen
file, (2) update `initialize` to use `_def`.
This ivar was passed to the thread function for the private state thread
- mostly because the equivalent variable was passed in in the original
version of the code. But it was never used, so I didn't notice that the
ivar equivalent wasn't being initialized. I'm going to keep the ivar
because it will be useful when debugging to easily see whether you are
on the main or secondary private state thread.
So in this change, I set m_is_secondary_thread properly in the
constructor, but remove passing it to the thread function since it
wasn't needed there.
We have a problem when some code on the private state thread needs to
run an expression. The private state thread is the one that fetches
"raw" events from the Process Plugin and decides how to handle them. But
if the private state thread needs to fetch the processed events to drive
running an expression, it can't also be the thread processing the raw
events.
We solve this by swapping in a modal private state thread just to handle
the events from the expression evaluation. That worked until you could
cause the expression evaluation to happen from Python, because then it
wasn't just the fetching of events that matter, but also the state of
the process and the state of the runlocks. The modal private state
thread is really a modal version of the thread and its associated state.
This patch gathers all the relevant control parameters into a structure
which we can swap in and out when needed.
It also adds a test using the new "was_hit" breakpoint resolver
affordance, which, since it acts as an asynchronous breakpoint callback,
gets run on the private state thread, showing that with the change we
can call expressions in the `was_hit` callback without problems.
I noticed that Module::GetMemoryObjectFile populates a Status object
upon error but it's effectively dropped on the floor. Instead, the
clients can report the error as desired.
At the moment, all clients are either (1) consuming the error because
it's only trying to find a module, or (2) log the error and bail out
early. I tried to preserve existing behavior as faithfully as possible.
This commit uses Process::ReadMemoryRanges to create an efficient method
for reading multiple strings at once. This method works like the
single-string version, reading 256 bytes at a time, but instead doing it
for _every_ string requested at the same time.
Both `Target::ReadSignedIntegerFromMemory()` and
`Process::ReadSignedIntegerFromMemory()` internally created an unsigned
scalar, so extending the value later did not duplicate the sign bit.
We weren't setting `m_should_detach` when going through the
`DoConnectRemote` code path. This meant that when you would attaches to
a remote process with `gdb-remote <port>` and use Ctrl+D, it would kill
the process instead of detach from it.
rdar://156111423
https://github.com/llvm/llvm-project/pull/165996 is adding a Clang
dependency to Target because we're moving some of the functionality of
the VerboseTrapFrameRecognizer into libClang. To avoid adding this
dependency this patch moves VerboseTrapFrameRecognizer into the
CPPLanguageRuntime. Most of the frame recognizers already live in the
various runtime plugins.
An alternative discussed was to create a common `CLanguageRuntime` whose
currently sole responsibility was to register the
`VerboseTrapFrameRecognizer` and `AssertStackFrameRecognizer`. The main
issue I ran into here was frame recognizers aren't uniqued in the
target. Currently this only manifests when re-running a target, which
re-triggers all the recognizer registration (added a test with a FIXME
for this). If we had a common `CLanguageRuntime` that
`CPPLanguageRuntime` and `ObjCLanguageRuntime` inherited from, I didn't
find a great way to avoid registering the recognizer multiple times. We
can't just call_once on it because we do want the recognisers to be
re-registered for new targets in the same debugger session. If the
recognisers were stored in something like a UniqueVector in the Target,
then we wouldn't have that issue. But currently that's not the case, and
it would take a bit of refactoring to de-dupe the recognisers.
There may very well be solutions I haven't considered, but all the
things I've tried so far I wasn't very happy with. So in the end I just
moved this to the C++ runtime for now in order to unblock
https://github.com/llvm/llvm-project/pull/165996.
The C++ language runtime is always available (even for C targets) if the
C++ language plugin is available. Which it should also be unless someone
is using an LLDB with the C++ plugin compiled out. But at that point
numerous things wouldn't work when even debugging just C.
This commit introduces a base-class implementation for a method that
reads memory from multiple ranges at once. This implementation simply
calls the underlying `ReadMemoryFromInferior` method on each requested
range, intentionally bypassing the memory caching mechanism (though this
may be easily changed in the future).
`Process` implementations that can be perform this operation more
efficiently - e.g. with the MultiMemPacket described in [1] - are
expected to override this method.
As an example, this commit changes AppleObjCClassDescriptorV2 to use the
new API.
Note about the API
------------------
In the RFC, we discussed having the API return some kind of class
`ReadMemoryRangesResult`. However, while writing such a class, it became
clear that it was merely wrapping a vector, without providing anything
useful. For example, this class:
```
struct ReadMemoryRangesResult {
ReadMemoryRangesResult(
llvm::SmallVector<llvm::MutableArrayRef<uint8_t>> ranges)
: ranges(std::move(ranges)) {}
llvm::ArrayRef<llvm::MutableArrayRef<uint8_t>> getRanges() const {
return ranges;
}
private:
llvm::SmallVector<llvm::MutableArrayRef<uint8_t>> ranges;
};
```
As can be seen in the added test and in the added use-case
(AppleObjCClassDescriptorV2), users of this API will just iterate over
the vector of memory buffers. So they want a return type that can be
iterated over, and the vector seems more natural than creating a new
class and defining iterators for it.
Likewise, in the RFC, we discussed wrapping the result into an
`Expected`. Upon experimenting with the code, this feels like it limits
what the API is able to do as the base class implementation never needs
to fail the entire result, it's the individual reads that may fail and
this is expressed through a zero-length result. Any derived classes
overriding `ReadMemoryRanges` should also never produce a top level
failure: if they did, they can just fall back to the base class
implementation, which would produce a better result.
The choice of having the caller allocate a buffer and pass it to
`Process::ReadMemoryRanges` is done mostly to follow conventions already
done in the Process class.
[1]:
https://discourse.llvm.org/t/rfc-a-new-vectorized-memory-read-packet/
The motivation for this patch is that
`StopInfo::GetSuggestedStackFrameIndex` would not take effect for
`InstrumentationRuntimeStopInfo` (which we plan to implement in
https://github.com/llvm/llvm-project/pull/133079). This was happening
because the instrumentation runtime plugins would run utility
expressions as part of the stop that would set the
`m_selected_frame_idx`. This means `SelectMostRelevantFrame` was never
called, and we would not be able to report the selected frame via the
`StopInfo` object.
This patch makes sure we clear the `m_selected_frame_idx` to allow
`GetSuggestedStackFrameIndex` to take effect, regardless of what the
frame recognizers choose to do.
This patch uses the "setting changed" callback to clear previously
cached stack frames when
`target.process.disable-language-runtime-unwindplans` is changed. This
is necessary so that changing the setting followed by a `backtrace`
command produces an accurate backtrace.
With this, a user can create a custom command like below in order to
quickly inspect a backtrace created without language plugins.
```
debugger.HandleCommand("settings set target.process.disable-language-runtime-unwindplans true")
debugger.HandleCommand("bt " + command)
debugger.HandleCommand("settings set target.process.disable-language-runtime-unwindplans false")
```
In the future, we may consider implementing this as an option to
`backtrace`. Currently, this process setting is the only way of
affecting the unwinder, and changing the process setting as part of the
backtrace implementation doesn't feel right.
There are no upstream users of this flag, so we cannot write a test for
it here.
This is an odd corner case of the use of scripts loaded from dSYM's - a
macOS only feature, which can load OS Plugins that re-present the thread
state of the program we attach to. If we find out about and load the
dSYM scripts when we discover a target in the course of attaching to it,
we can end up running the OS plugin before we've started up the private
state thread. However, the os_plugin in that case will be running before
we broadcast the stop event to the public event listener. So it should
formally use the private state and not the public state for the Python
code environment.
This patch says that if we have not yet started up the private state
thread, then any thread that is servicing events is doing so on behalf
of the private state machinery, and should see the private state, not
the public state.
Most of the patch is getting a test that will actually reproduce the
error. Only the test `test_python_os_plugin_remote` actually reproduced
the error. In `test_python_os_plugin` we actually do start up the
private state thread before handling the event. `test_python_os_plugin`
is there for completeness sake.
It's not necessary on posix platforms as of #126935 and it's ignored on
windows as of #138896. For both platforms, we have a better way of
inheriting FDs/HANDLEs.
This patch should be mostly obvious, but in one place, this patch
changes:
const auto &it = std::find(...)
to:
auto it = llvm::find(...)
We do not need to bind to a temporary with const ref.
This macro does not do what is described. The only thing it actually
does control is whether or not the process's memory cache gets flushed
when writing to an address. It does not override the setting
`target.process.disable-memory-cache`.
Instead of making it work as intended, I chose to remove the macro. I
don't see much value in being able to override the setting with a
preprocessor macro.
This reapplies https://github.com/llvm/llvm-project/pull/138892, which
was reverted in
5fb9dca14a
due to failures on windows.
Windows loads modules from the Process class, and it does that quite
early, and it kinda makes sense which is why I'm moving the clearing
code even earlier.
The original commit message was:
Minidump files contain explicit information about load addresses of
modules, so it can load them itself. This works on other platforms, but
fails on darwin because DynamicLoaderDarwin nukes the loaded module list
on initialization (which happens after the core file plugin has done its
work).
This used to work until
https://github.com/llvm/llvm-project/pull/109477, which enabled the
dynamic loader
plugins for minidump files in order to get them to provide access to
TLS.
Clearing the load list makes sense, but I think we could do it earlier
in the process, so that both Process and DynamicLoader plugins get a
chance to load modules. This patch does that by calling the function
early in the launch/attach/load core flows.
This fixes TestDynamicValue.py:test_from_core_file on darwin.
Follow-up to #138892 fixing breakage on windows. Calling
ClearAllLoadedSections earlier is necessary to avoid throwing out the
work done by the windows process plugin.
Minidump files contain explicit information about load addresses of
modules, so it can load them itself. This works on other platforms, but
fails on darwin because DynamicLoaderDarwin nukes the loaded module list
on initialization (which happens after the core file plugin has done its
work).
This used to work until #109477, which enabled the dynamic loader
plugins for minidump files in order to get them to provide access to
TLS.
Clearing the load list makes sense, but I think we could do it earlier
in the process, so that both Process and DynamicLoader plugins get a
chance to load modules. This patch does that by calling the function
early in the launch/attach/load core flows.
This fixes TestDynamicValue.py:test_from_core_file on darwin.
Custom regions in Process::GetUserSpecifiedCoreFileSaveRanges originally
used `FindEntryThatContains`. This made sense on my first attempt, but
what we really want are *intersecting* regions. This is so the user can
specify arbitrary memory, and if it's available we output it to the core
(Minidump or MachO).
This reverts commit daa4061d61216456baa83ab404e096200e327fb4.
Original PR https://github.com/llvm/llvm-project/pull/129092.
I have restricted the test to X86 Windows because it turns out the only
reason that `expr x.get()` would change m_memory_id is that on x86 we
have to write the return address to the stack in ABIWindows_X86_64::PrepareTrivialCall:
```
// Save return address onto the stack
if (!process_sp->WritePointerToMemory(sp, return_addr, error))
return false;
```
This is not required on AArch64 so m_memory_id was not changed:
```
(lldb) expr x.get()
(int) $0 = 0
(lldb) process status -d
Process 15316 stopped
* thread #1, stop reason = Exception 0x80000003 encountered at address 0x7ff764a31034
frame #0: 0x00007ff764a31038 TestProcessModificationIdOnExpr.cpp.tmp`main at TestProcessModificationIdOnExpr.cpp:35
32 __builtin_debugtrap();
33 __builtin_debugtrap();
34 return 0;
-> 35 }
36
37 // CHECK-LABEL: process status -d
38 // CHECK: m_stop_id: 2
ProcessModID:
m_stop_id: 3
m_last_natural_stop_id: 0
m_resume_id: 0
m_memory_id: 0
```
Really we should find a better way to force a memory write here, but
I can't think of one right now.
And a follow up warning fix.
This reverts commit 6aa963f780d63d4c8fa80de97dd79c932bc35f4e
and 2bff80f25d51e24d3c552e033a2863dd36ef648b.
This is failing on Windows on Arm: https://lab.llvm.org/buildbot/#/builders/141/builds/8375
Seems to produce the line the test wants but not in the right place.
Reverting while I investigate.
stop-hooks are supposed to trigger every time the process stops, but as
initially implemented they would only fire when control was returned to
the user. So for instance when a process was launched the stop hook
would only trigger when the process hit a breakpoint or crashed.
However, it would be really useful to be able to trigger a stop hook
when lldb first gains control over the process. One way to do that would
be to implement general "target lifecycle events" and then send process
created events that users could bind actions to.
OTOH, extending the stop hooks to fire when lldb first gains control
over the process is a pretty natural extension to the notion of a stop
hook. So this patch takes the shorter route to that ability by making
stop-hooks fire when lldb first gains control over the process.
I also added the ability to specify whether to trigger the stop hook "on
gaining control". I'm on the fence about whether to set the default to
be "trigger on gaining control" or "don't trigger on gaining control".
Since I think it's a generally useful feature, I've set the default to
"trigger on gaining control".
This change adds a setting `target.process.track-memory-cache-changes`.
Disabling this setting prevents invalidating and updating values in
`ValueObject::UpdateValueIfNeeded` when only "internal" debugger memory
is updated. Writing to "internal" debugger memory happens when, for
instance, expressions are evaluated by visualizers (pretty printers).
One of the examples when cache invalidation has a particularly heavy
impact is visualizations of some collections: in some collections
getting collection size is an expensive operation (it requires traversal
of the collection).
At the same time evaluating user expression with side effects (visible
to target, not only to debugger) will still bump memory ID because:
- If expression is evaluated via interpreter: it will cause write to
"non-internal" memory
- If expression is JIT-compiled: then to call the function LLDB will
write to "non-internal" stack memory
The downside of disabled `target.process.track-memory-cache-changes`
setting is that convenience variables won't reevaluate synthetic
children automatically.
---------
Co-authored-by: Mikhail Zakharov <mikhail.zakharov@jetbrains.com>
I traced the issue reported by Caroline and Pavel in #134757 back to the
call to ProcessRunLock::TrySetRunning. When that fails, we get a
somewhat misleading error message:
> process resume at entry point failed: Resume request failed - process
still running.
This is incorrect: the problem was not that the process was in a running
state, but rather that the RunLock was being held by another thread
(i.e. the Statusline). TrySetRunning would return false in both cases
and the call site only accounted for the former.
Besides the odd semantics, the current implementation is inherently
race-y and I believe incorrect. If someone is holding the RunLock, the
resume call should block, rather than give up, and with the lock held,
switch the running state and report the old running state.
This patch removes ProcessRunLock::TrySetRunning and updates all callers
to use ProcessRunLock::SetRunning instead. To support that,
ProcessRunLock::SetRunning (and ProcessRunLock::SetStopped, for
consistency) now report whether the process was stopped or running
respectively. Previously, both methods returned true unconditionally.
The old code has been around pretty much pretty much forever, there's
nothing in the git history to indicate that this was done purposely to
solve a particular issue. I've tested this on both Linux and macOS and
confirmed that this solves the statusline issue.
A big thank you to Jim for reviewing my proposed solution offline and
trying to poke holes in it.
I recently received an internal error report that LLDB was OOM'ing when
creating a Minidump. In my 64b refactor we made a decision to acquire
buffers the size of the largest memory region so we could read all of
the contents in one call. This made error handling very simple (and
simpler coding for me!) but had the trade off of large allocations if
huge pages were enabled.
This patch is one I've had on the back burner for awhile, but we can
read and write the Minidump memory sections in discrete chunks which we
already do for writing to disk.
I had to refactor the error handling a bit, but it remains the same. We
make a best effort attempt to read as much of the memory region as
possible, but fail immediately if we receive an error writing to disk. I
did not add new tests for this because our existing test suite is quite
good, but I did manually verify a few Minidumps couldn't read beyond the
red_zone.
```
(lldb) reg read $sp
rsp = 0x00007fffffffc3b0
(lldb) p/x 0x00007fffffffc3b0 - 128
(long) 0x00007fffffffc330
(lldb) memory read 0x00007fffffffc330
0x7fffffffc330: 60 c3 ff ff ff 7f 00 00 60 cd ff ff ff 7f 00 00 `.......`.......
0x7fffffffc340: 60 c3 ff ff ff 7f 00 00 65 e6 26 00 00 00 00 00 `.......e.&.....
(lldb) memory read 0x00007fffffffc329
error: could not parse memory info (Success!)
```
I'm not sure how to quantify the memory improvement other than we would
allocate the largest size regardless of the size. So a 2gb unreadable
region would cause a 2gb allocation even if we were reading 4096 kb. Now
we will take the range size or the max chunk size of 128 mb.
This NFC patch simplifies the main loop in HandleProcessStateChanged
event by moving duplicated code into the StopInfo class, also allowing
StopInfo subclasses to override behavior.
More specifically, two functions are created:
* ShouldShow: should a Thread with such StopInfo should be printed when
the debugger stops? Currently, no StopInfo subclasses override this, but
a subsequent patch will fix a bug by making StopInfoBreakpoint check
whether the breakpoint is internal.
* ShouldSelect: should a Thread with such a StopInfo be selected? This
is currently overridden by StopInfoUnixSignal but will, in the future,
be overridden by StopInfoBreakpoint.
When using `SBFrame::EvaluateExpression` on a frame that's not the
currently selected frame, we would sometimes run into errors such as:
```
error: error: The context has changed before we could JIT the expression!
error: errored out in DoExecute, couldn't PrepareToExecuteJITExpression
```
During expression parsing, we call `RunStaticInitializers`. On our
internal fork this happens quite frequently because any usage of, e.g.,
function pointers, will inject ptrauth fixup code into the expression.
The static initializers are run using `RunThreadPlan`. The
`ExecutionContext::m_frame_sp` going into the `RunThreadPlan` is the
`SBFrame` that we called `EvaluateExpression` on. LLDB then tries to
save this frame to restore it after the thread-plan ran (the restore
occurs by unconditionally overwriting whatever is in
`ExecutionContext::m_frame_sp`). However, if the `selected_frame_sp` is
not the same as the `SBFrame`, then `RunThreadPlan` would set the
`ExecutionContext`'s frame to a different frame than what we started
with. When we `PrepareToExecuteJITExpression`, LLDB checks whether the
`ExecutionContext` frame changed from when we initially
`EvaluateExpression`, and if did, bails out with the error above.
One such test-case is attached. This currently passes regardless of the
fix because our ptrauth static initializers code isn't upstream yet. But
the plan is to upstream it soon.
This patch addresses the issue by saving/restoring the frame of the
incoming `ExecutionContext`, if such frame exists. Otherwise, fall back
to using the selected frame.
rdar://147456589
SetExitStatus can be called the second time when we reap the debug
server process. This shouldn't be interesting as at that point, we've
already told everyone that the process has exited.
I believe/hope this will also help with sporadic shutdown crashes that
have cropped up recently. They happen because the debug server is
monitored from a detached thread, so this code can be called after main
returns (and starts destroying everything). This isn't a real fix for
that though, as the situation can still happen (it's just that it
usually happens after the exit status has already been set). I think the
real fix for that is to make sure these threads terminate before we
start shutting everything down.
This reverts commit
87b7f63a11,
reapplying
7e66cf74fb
with a small (and probably temporary)
change to generate more debug info to help with diagnosing buildbot
issues.
The main motivation for this was the inconsistency in handling of
partial reads/writes between the windows and posix implementations
(windows was returning partial reads, posix was trying to fill the
buffer completely). I settle on the windows implementation, as that's
the more common behavior, and the "eager" version can be implemented on
top of that (in most cases, it isn't necessary, since we're writing just
a single byte).
Since this also required auditing the callers to make sure they're
handling partial reads/writes correctly, I used the opportunity to
modernize the function signatures as a forcing function. They now use
the `Timeout` class (basically an `optional<duration>`) to support both
polls (timeout=0) and blocking (timeout=nullopt) operations in a single
function, and use an `Expected` instead of a by-ref result to return the
number of bytes read/written.
As a drive-by, I also fix a problem with the windows implementation
where we were rounding the timeout value down, which meant that calls
could time out slightly sooner than expected.
