mirrors/cdt
1
0
Fork 0
mirror of https://github.com/eclipse-cdt/cdt synced 2025-08-03 22:35:43 +02:00
Code Issues Wiki Activity

Bug 458091 - Debug frames are missing or not shown in some cases

Browse source 
- one case where debugger fails to return stack depth, but returns
correct stack frames
- second case where debugger fails to return stack frames when asked
without limit, but can return a specific frame

Both of these are workaround for GDB bugs where it fails to return data
consistently

Change-Id: I4fb0d4e850fbce6adc655849ec3f73c85adb1bdc
Signed-off-by: Alena Laskavaia <elaskavaia.cdt@gmail.com>
This commit is contained in:
Alena Laskavaia 2015-01-21 16:27:40 -05:00 • committed by Gerrit Code Review @ Eclipse.org
parent b7829381f2
commit 5a7e047ab5
1 changed files with 85 additions and 61 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

146
dsf-gdb/org.eclipse.cdt.dsf.gdb/src/org/eclipse/cdt/dsf/mi/service/MIStack.java
View file

@ -493,56 +493,53 @@ public class MIStack extends AbstractDsfService
protected MIFrame getMIFrame() { return fFrameDataCacheInfo.getMIFrames()[fFrameIndex]; }
}
fMICommandCache.execute(
fCommandFactory.createMIStackListFrames(execDmc),
new DataRequestMonitor<MIStackListFramesInfo>(getExecutor(), rm) {
@Override
protected void handleSuccess() {
// Find the index to the correct MI frame object.
int idx = findFrameIndex(getData().getMIFrames(), miFrameDmc.fLevel);
if (idx == -1) {
rm.setStatus(new Status(IStatus.ERROR, GdbPlugin.PLUGIN_ID, INVALID_HANDLE, "Invalid frame " + frameDmc, null)); //$NON-NLS-1$
rm.done();
return;
}
// Create the data object.
rm.setData(new FrameDataFromMIStackFrameListInfo(getData(), idx));
rm.done();
}
@Override
protected void handleError() {
// We're seeing gdb in some cases fail when it's
// being asked for the stack depth or stack
// frames, but the same command succeeds if
// the request is limited to one frame. So try
// again with a limit of 1. It's better to show
// just one frame than none at all
if (miFrameDmc.fLevel == 0) {
fMICommandCache.execute(
fCommandFactory.createMIStackListFrames(execDmc, 0, 0),
new DataRequestMonitor<MIStackListFramesInfo>(getExecutor(), rm) {
@Override
protected void handleSuccess() {
// Find the index to the correct MI frame object.
int idx = findFrameIndex(getData().getMIFrames(), miFrameDmc.fLevel);
if (idx == -1) {
rm.setStatus(new Status(IStatus.ERROR, GdbPlugin.PLUGIN_ID, INVALID_HANDLE, "Invalid frame " + frameDmc, null)); //$NON-NLS-1$
rm.done();
return;
}
// Create the data object.
rm.setData(new FrameDataFromMIStackFrameListInfo(getData(), idx));
rm.done();
}
});
} else {
super.handleError();
}
}
});
fMICommandCache.execute(
fCommandFactory.createMIStackListFrames(execDmc),
new DataRequestMonitor<MIStackListFramesInfo>(getExecutor(), rm) {
@Override
protected void handleSuccess() {
// Find the index to the correct MI frame object.
int idx = findFrameIndex(getData().getMIFrames(), miFrameDmc.fLevel);
if (idx == -1) {
rm.setStatus(new Status(IStatus.ERROR, GdbPlugin.PLUGIN_ID, INVALID_HANDLE,
"Invalid frame " + frameDmc, null)); //$NON-NLS-1$
rm.done();
return;
}
// Create the data object.
rm.setData(new FrameDataFromMIStackFrameListInfo(getData(), idx));
rm.done();
}
@Override
protected void handleError() {
// We're seeing gdb in some cases fail when it's
// being asked for the stack depth or stack
// frames, but the same command succeeds if
// the request is limited to one frame. So try
// again for a specific frame. It's better to show
// just one frame than none at all. Since it is only happen on error
// this should not contribute much to increased traffic
fMICommandCache.execute(
fCommandFactory.createMIStackListFrames(execDmc, miFrameDmc.fLevel, miFrameDmc.fLevel),
new DataRequestMonitor<MIStackListFramesInfo>(getExecutor(), rm) {
@Override
protected void handleSuccess() {
// Find the index to the correct MI frame object.
int idx = findFrameIndex(getData().getMIFrames(), miFrameDmc.fLevel);
if (idx == -1) {
rm.setStatus(new Status(IStatus.ERROR, GdbPlugin.PLUGIN_ID, INVALID_HANDLE,
"Invalid frame " + frameDmc, null)); //$NON-NLS-1$
rm.done();
return;
}
// Create the data object.
rm.setData(new FrameDataFromMIStackFrameListInfo(getData(), idx));
rm.done();
}
});
}
});
}
@Override
@ -944,17 +941,44 @@ public class MIStack extends AbstractDsfService
rm.done();
} else {
// We're seeing gdb in some cases fail when it's
// being asked for the stack depth or stack
// frames, but the same command succeeds if
// the request is limited to one frame. So try
// again with a limit of 1. It's better to show
// just one frame than none at all
if (maxDepth != 1) {
getStackDepth(dmc, 1, rm);
}
else {
super.handleError();
}
// being asked for the stack depth but stack frames command succeeds
// it seems like an overkill but it will cached and ui later will ask for it anyway
ICommand<MIStackListFramesInfo> listFramesCommand;
if (maxDepth <= 0)
listFramesCommand = fCommandFactory.createMIStackListFrames(execDmc);
else
listFramesCommand = fCommandFactory.createMIStackListFrames(execDmc, 0, maxDepth - 1);
fMICommandCache.execute(
listFramesCommand,
new DataRequestMonitor<MIStackListFramesInfo>(getExecutor(), rm) {
@Override
protected void handleSuccess() {
try {
// Find the maximum level in returned frames
MIFrame[] miFrames = getData().getMIFrames();
int level = 0;
for (MIFrame miFrame : miFrames) {
if (miFrame.getLevel() > level)
level = miFrame.getLevel();
}
// Create the data object. Depth is +1 of maximum frame level
int depth = level + 1;
fStackDepthCache.put(execDmc.getThreadId(), new StackDepthInfo(maxDepth, depth));
rm.setData(depth);
} finally {
rm.done(); // we have to close monitor no matter what
}
}
@Override
protected void handleError() {
// There is no point asking for stack-depth with limit of 1, lets just assume it is
// at least 1, worst case it will show error or no data on the first frame
rm.setData(1);
rm.done();
};
}
);
}
}
});