jingham added a comment.
The debug info is parsed only when needed. So for breakpoints, it will get
parsed when the breakpoint is set. That's one good use of "break set -s
" BTW, it will limit the debug info read in to just the library
you expect to find the symbol in.
https://reviews.llvm.
jingham added a comment.
If you want to avoid loading dependent libraries, you can use the "-d" flag to
"target create". This is sometimes handy, but shouldn't be the default
behavior.
https://reviews.llvm.org/D43419
___
lldb-commits mailing list
zturner added a comment.
I guess on the other hand, it's reasonable to assume that if you've set a
breakpoint somewhere, you're more likely than not to need it since you probably
had a reason for setting it. Is the debug info parsed when the executable is
loaded, or when the breakpoint is set?
zturner added a comment.
I guess one advantage to delaying it is that the debug info for the dynamic
library could be large and slow to parse, and you don't know if you're even
going to need it until you hit the breakpoint. So by delaying the resolution
even to File address, you postpone parsi
jingham added a comment.
It is really handy to see the pre-load libraries in lldb, it means you can do
"source list" to make sure the breakpoints you are planning to set are where
you expect them to be, and tab completion can help you set breakpoints pre-run,
etc. So if you can do this, your u
zturner added a comment.
I'm not sure how hard it would be, but it would be better if we could do the
same thing on Windows, for consistency of behavior. In principle it's
straightforward, just scan through the IAT and load all of the imported
modules. Probably the DynamicLoaderWindows plugin
amccarth added a comment.
OK, so my fix is too simplistic. I'll have to (1) make LLDB on Windows also
attempt to pre-load the DLLs, (2) make the test use platform-specific
expectations, or (3) use a looser expectation. I'll give it some thought over
the weekend.
https://reviews.llvm.org/D43
jasonmolenda added a comment.
Yeah, it has a location because we're resolved it to a File address
(section+offset), and when the dylib actually gets loaded we'll resolve that to
a load address and insert the breakpoint instruction.
https://reviews.llvm.org/D43419
___
zturner added a comment.
In https://reviews.llvm.org/D43419#1011045, @jasonmolenda wrote:
> On Darwin we load all the libraries that the binary links against
> pre-execution, if possible. So I see:
>
> % lldb a.out
> (lldb) ima li libfoo.dylib
> [ 0] 35C5FE62-5327-3335-BBCF-5369CB07D1D6 0x00
jasonmolenda added a comment.
On Darwin we load all the libraries that the binary links against
pre-execution, if possible. So I see:
% lldb a.out
(lldb) ima li libfoo.dylib
[ 0] 35C5FE62-5327-3335-BBCF-5369CB07D1D6 0x
/Volumes/ssdhome/jmolenda/k/svn/lldb/packages/Python/lldbs
zturner added a reviewer: labath.
zturner added a comment.
What's the behavior on other platforms? When you set a breakpoint in a shared
library before you've run the program, shouldn't it still be unresolved, in
which case this test should have failed on those platforms too?
https://reviews.
amccarth created this revision.
amccarth added a reviewer: jasonmolenda.
Herald added a subscriber: sanjoy.
This test was failing on Windows because it expected the breakpoint in the
dynamic library to be resolved before the process is launched. Since the DLL
isn't loaded until the process is lau
jingham accepted this revision.
jingham added a comment.
Excellent, thanks.
https://reviews.llvm.org/D4
___
lldb-commits mailing list
lldb-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
davide accepted this revision.
davide added a comment.
This revision is now accepted and ready to land.
lgtm
https://reviews.llvm.org/D43376
___
lldb-commits mailing list
lldb-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/lld
clayborg accepted this revision.
clayborg added a comment.
Looks good. Self describing and concise.
https://reviews.llvm.org/D4
___
lldb-commits mailing list
lldb-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
Author: labath
Date: Fri Feb 16 04:57:35 2018
New Revision: 325340
URL: http://llvm.org/viewvc/llvm-project?rev=325340&view=rev
Log:
Shorten socket names in TestPlatformProcessConnect
The test was generating long unix socket names, and the addition of a
new folder in the previous patch pushed it
Author: jdevlieghere
Date: Fri Feb 16 04:33:10 2018
New Revision: 325339
URL: http://llvm.org/viewvc/llvm-project?rev=325339&view=rev
Log:
Re-enable lang/objc/modules/TestObjCModules
The reason this test was disabled is no longer relevant. However, it
didn't turn into an unexpected success becaus
Author: labath
Date: Fri Feb 16 03:39:38 2018
New Revision: 325334
URL: http://llvm.org/viewvc/llvm-project?rev=325334&view=rev
Log:
Fix paralelization of remote tests
Since we now can run multiple tests from the same directory at once, we
need to include the test name in the remote test director
labath created this revision.
labath added reviewers: eugene, davide.
The issue was that we were parsing the registers into 64-bit integers
and the calling swapByteOrder without regard for the actual size of the
register. This switches the test to use the RegisterValue class which
tracks the regis
labath abandoned this revision.
labath added a comment.
It looks like we'll do something like https://reviews.llvm.org/D4 instead...
https://reviews.llvm.org/D43292
___
lldb-commits mailing list
lldb-commits@lists.llvm.org
http://lists.llvm.org/
Author: labath
Date: Fri Feb 16 01:29:41 2018
New Revision: 325324
URL: http://llvm.org/viewvc/llvm-project?rev=325324&view=rev
Log:
[lldb][PPC64] Fixed vector and struct return value
Summary:
The PowerPC64 ABI plugin was modified to:
- properly handle vector type return values
- implement suppo
This revision was automatically updated to reflect the committed changes.
Closed by commit rL325322: [dosep] Run tests in a more parallel fashion
(authored by labath, committed by ).
Herald added a subscriber: llvm-commits.
Repository:
rL LLVM
https://reviews.llvm.org/D43335
Files:
lldb/tru
Author: labath
Date: Fri Feb 16 01:21:11 2018
New Revision: 325322
URL: http://llvm.org/viewvc/llvm-project?rev=325322&view=rev
Log:
[dosep] Run tests in a more parallel fashion
Summary:
Due to in-tree builds, we were parallelizing the tests at the directory
level. Now that the tests are built ou
labath updated this revision to Diff 134564.
labath added a comment.
I've updated the format to include a description of the item. I don't we need to
worry about the expensiveness of this data structure too much, as it should
never be used in a hot loop (the only use case I can think of is queryin
24 matches
Mail list logo
