hubert.reinterpretcast marked 6 inline comments as done.
hubert.reinterpretcast added a comment.
Herald added a subscriber: jsji.
Thanks @xingxue for the review. I will update for the error handling before
committing.
================
Comment at: tools/libclang/CIndexer.cpp:61
+ if (errno != ENOMEM)
+ llvm_unreachable("Encountered an unexpected loadquery() failure");
+
----------------
hubert.reinterpretcast wrote:
> Based on available documentation, this situation should not occur (and thus I
> believe `llvm_unreachable` is appropriate).
I'll leave this one as `llvm_unreachable`, either the failure is resolved on a
later call or the loop will terminate when the amount of memory to request
(checked below) is too large.
================
Comment at: tools/libclang/CIndexer.cpp:65
+ if ((BufSize & ~((-1u) >> 1u)) != 0u)
+ llvm_unreachable("BufSize needed for loadquery() too large");
+
----------------
hubert.reinterpretcast wrote:
> This situation is not impossible, but highly improbable. This is a
> non-programmatic error, and the Programmer's Manual appears to recommend the
> use of `report_fatal_error`. Some additional guidance would be appreciated.
I will change this to `report_fatal_error`.
================
Comment at: tools/libclang/CIndexer.cpp:84
+ if (CurInfo->ldinfo_next == 0u)
+ llvm_unreachable("Cannot locate entry point in the loadquery() results");
+ CurInfo = reinterpret_cast<ld_xinfo *>(reinterpret_cast<char *>(CurInfo) +
----------------
hubert.reinterpretcast wrote:
> This is also supposed to not happen.
I will change this to `report_fatal_error`. Infinite loops are unfriendly.
Repository:
rC Clang
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D59233/new/
https://reviews.llvm.org/D59233
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
