- User Since
- Apr 16 2019, 12:23 PM (131 w, 3 d)
Nov 21 2019
Oct 30 2019
Oct 29 2019
Moving to a single config, use-g-packet-for-reading, that forces the use of 'g' packets for reading, but does not force 'G' for writing. The latter only ends up being used if 'p' packets aren't supported (it assumes that 'P' also will not).
Oct 28 2019
Oct 24 2019
Have you had a chance to take a look at the last revision of this?
Oct 11 2019
Properly separating MPX fix from this change (I guess)
Moving MPX fix to a parent change
Oct 9 2019
Dissociating the use of g/G packets and trying to address the AVX/MPX offset bug
Oct 7 2019
Thank you for looking into this, @labath
Sep 16 2019
Sep 13 2019
Sep 9 2019
Maybe we could use the previous workaround until we find something better?
Falling back to the previous workaround to test EnsureAllDIEsInDeclContextHaveBeenParsed.
Sep 6 2019
Sep 5 2019
Sep 4 2019
Migrating to EnsureAllDIEsInDeclContextHaveBeenParsed
Sep 3 2019
Renaming the test
Aug 30 2019
Aug 22 2019
Sounds good, thank you!
Aug 20 2019
I'm fixing the style issues pointed by @labath, but I acknowledge the flakiness of this test the way it is right now and am open to deleting it. I can also try to spend some time trying to implement it in a more proper manner. It's your call, @clayborg.
Aug 19 2019
I tried to write a unit test following @labath's suggestion of creating a test class that inherits from DWARFASTParserClang. Please let me know what you guys think.
Aug 17 2019
Aug 16 2019
This is a follow up on the investigation I mentioned in http://lists.llvm.org/pipermail/lldb-dev/2019-August/015324.html.
Please let me know if you guys think this makes sense. Thanks!
Jul 30 2019
Oh, sorry about that. I was relying on ninja check, which runs okay for me.
Jul 29 2019
Adding support for tablegen generated properties introduced in https://reviews.llvm.org/D65185
Jul 26 2019
Could anyone please merge this for me?
Jul 25 2019
Adds a test to make sure the client is using 'g' by default.
Jul 10 2019
Jul 9 2019
This rebases the change, rename *try_to_use_g_packet* to *use_g_packet*, defaults the setting to true, and modifies the condition of the read_all_registers_at_once bool in ThreadGDBRemote::CreateRegisterContextForFrame to guarantee the server supports 'g' packet.
Jun 25 2019
Just checking the status of this change. Are we still considering enabling 'g' packets unconditionally, or are we better off using the setting for safety?
Jun 7 2019
Jun 5 2019
May 29 2019
May 28 2019
Remove redundant cast from GDBRemoteCommunicationServerLLGS::Handle_g and add annotation to TestGdbRemoteGPacket::g_returns_correct_data so it gets skipped if the running architecture isn't x86_64
May 27 2019
Add inferior to to set the values of a few registers to known values so we can verify the 'g' packet looks good. Also, start testing with/without thread suffix .
May 22 2019
Thank you so much for the feedback!
Making this revision be responsible only for adding the 'g' packet handler and tests. Also, I'm starting to implement a stronger test to verify the actual content of the packets returned by the new handler.