My personal website source code
Log | Files | Refs | Submodules | README | LICENSE

commit d4d2925877658f875237a0201a038934d42da1dd
parent 324b7420fc7544ee3c133f367a0157aa8d6885fc
Author: Luís Ferreira <[email protected]>
Date:   Fri, 17 Dec 2021 00:38:30 +0000

posts: Add 'SAOC LLDB D integration: 13th Weekly Update'

Signed-off-by: Luís Ferreira <[email protected]>

Mcontent/posts/ | 3++-
Acontent/posts/ | 58++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
2 files changed, 60 insertions(+), 1 deletion(-)

diff --git a/content/posts/ b/content/posts/ @@ -68,4 +68,5 @@ You can also read this on the D programming language forum, [here]([email protected]), and discuss there! -Read about the [previous week](../d-saoc-2021-11/). +Read about the [previous week](../d-saoc-2021-11/) and the [next +week](../d-saoc-2021-13/). diff --git a/content/posts/ b/content/posts/ @@ -0,0 +1,58 @@ +--- +title: 'SAOC LLDB D integration: 13th Weekly Update' +date: '2021-12-16T23:34:00+01:00' +tags: ['saoc', 'saoc2021', 'dlang', 'llvm', 'lldb', 'debug', 'debugging', 'dwarf'] +description: "This post describes what I've done on the 13th week of the +Symmetry Autumn of Code 2021, including some bug fixes and trivial patches." +--- + +Hi D community! + +I'm here again, to describe what I've done during the thirteenth week of +Symmetry Autumn of Code. This week as a bit atypical as I got infected with +COVID-19 and this made me feel exhausted both physically and mentally. + +## Working approach to fetch basic types + +After digging a bit about the problem I found out there is no problem +specifically in Release builds, and a simple function called `Verify` is there +to double check if the type is correct for Debug builds. After fixing that +function, the LLDB can now dump some simple output about global TLS and non-TLS +variables: + +``` +(lldb) ta v +Global variables for app.d in app: = <No TLS data currently exists for this thread.> + +app.ptr = <could not resolve type> +app.foobar = +``` + +For now it only recognizes boolean types, but it can't show up the value since +LLDB doesn't know a way to dump it correctly. + +You can try this version yourself using +[this]( +branch. + +## Decoupling and trivial patches + +I pushed a patch to decouple DWARF to LLDB encoding logic from Clang DWARF AST +Parser, along with a trivial fix about duplicate map assignment: + +- +- + +## What is next? + +I already started working on some helpers to fetch the LLDB Format, Encoding +and BasicType, along with other information about a type to properly display +them. I'm not sure how Milestone 4 is going to work, as I underestimated +the support for custom expessions. + +You can also read this on the D programming language forum, +[here]([email protected]), +and discuss there! + +Read about the [previous week](../d-saoc-2021-12/).