diff options
author | Felipe de Azevedo Piovezan <fpiovezan@apple.com> | 2025-10-14 14:02:46 -0700 |
---|---|---|
committer | GitHub <noreply@github.com> | 2025-10-14 14:02:46 -0700 |
commit | bed17c03fee09eabbd35eca3a8829f913a374424 (patch) | |
tree | 53d75f594427054e6b4077bbb724fa59d335c34d /llvm/lib/Support/UnicodeNameToCodepoint.cpp | |
parent | 93185ea6fe0ad4a26e46635c190f1a8d7bf1933c (diff) | |
download | llvm-bed17c03fee09eabbd35eca3a8829f913a374424.zip llvm-bed17c03fee09eabbd35eca3a8829f913a374424.tar.gz llvm-bed17c03fee09eabbd35eca3a8829f913a374424.tar.bz2 |
[lldb][NFCI] Refactor AppleObjCClassDescriptorV2 method_t parsing functions (#163291)
This rewrites ClassDescriptorV2::method_t::Read (and the loop calling
that function) in an NFCI way to perform a couple of things:
1. Cleanup code with indirect style. For example, the old loop would
have default-constructor a `unique_ptr<method_t>`, which was *reused* on
every iteration of the loop. It called `method_t::Read` on each
iteration, and never checked the return value prior to invoking the
callback. In other words, if `Read` failed, the callback was called on
random values.
2. Exposed memory reads that could benefit from the MultiMemoryRead
packet proposed in [1].
[1]:
https://discourse.llvm.org/t/rfc-a-new-vectorized-memory-read-packet/88441
Diffstat (limited to 'llvm/lib/Support/UnicodeNameToCodepoint.cpp')
0 files changed, 0 insertions, 0 deletions