<div dir="ltr"><div dir="ltr">On Fri, Mar 27, 2020 at 3:23 AM Simon Pilgrim <<a href="mailto:llvm-dev@redking.me.uk">llvm-dev@redking.me.uk</a>> wrote:<br></div><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<div>
<p>Yes it was a clang static analyzer report listed on
<a href="https://llvm.org/reports/scan-build/" target="_blank">https://llvm.org/reports/scan-build/</a></p>
<p>I'd include the specific link in the commit message but they
don't persist very long.<br></p></div></blockquote><div>No worries - clearly I was a bit late reviewing these, so the links probably would've been dead. Maybe copy/pasting the diagnostics into the commit message might be handy, though? <br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div><p>
</p>
<p>BTW - I keep missing your feedback emails as they get lost in the
firehose that is llvm-commits - could you use phabricator instead
to discuss particular commits please?</p></div></blockquote><div>I know llvm-commits is a bit of a busy place (hence why it's taken me 3 months to get around to reviewing some things... ) - but even if I send you feedback that way, other people might not, so it seems important to have you mail setup so such messages don't get lost (using gmail I have anything where I am on the "to" line sent to my inbox and starred separately, for instance). People might be reporting urgent things this way - like post-commit breakages and other discussions & they can come from a variety of people who won't all be using Phabricator to provide that feedback.<br><br>- Dave</div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div><p>Cheers, Simon.</p>
<div>On 27/03/2020 01:36, David Blaikie
wrote:<br>
</div>
<blockquote type="cite">
<div dir="ltr">Was this for the Clang Static Analyzer? Any idea
what parts the analyzer couldn't understand?<br>
<br>
I assume it has to assume that function returns might already be
non-null (eg: T1B probably isn't "might be null" &
dereferencing it wouldn't be a problem? Or is any pointer return
"Maybe null" in which case maybe just asserting T1B and T2B are
non-null as soon as they're returned might be more
suitable/closer to the source of the divergence?)</div>
<br>
<div class="gmail_quote">
<div dir="ltr" class="gmail_attr">On Fri, Jan 10, 2020 at 3:11
AM Simon Pilgrim via llvm-commits <<a href="mailto:llvm-commits@lists.llvm.org" target="_blank">llvm-commits@lists.llvm.org</a>>
wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><br>
Author: Simon Pilgrim<br>
Date: 2020-01-10T11:10:42Z<br>
New Revision: 870f6917936fdb8050be3ca3c67d9259390c4326<br>
<br>
URL: <a href="https://github.com/llvm/llvm-project/commit/870f6917936fdb8050be3ca3c67d9259390c4326" rel="noreferrer" target="_blank">https://github.com/llvm/llvm-project/commit/870f6917936fdb8050be3ca3c67d9259390c4326</a><br>
DIFF: <a href="https://github.com/llvm/llvm-project/commit/870f6917936fdb8050be3ca3c67d9259390c4326.diff" rel="noreferrer" target="_blank">https://github.com/llvm/llvm-project/commit/870f6917936fdb8050be3ca3c67d9259390c4326.diff</a><br>
<br>
LOG: Fix "pointer is null" static analyzer warnings. NFCI.<br>
<br>
Assert that the pointers are non-null before dereferencing
them.<br>
<br>
Added: <br>
<br>
<br>
Modified: <br>
llvm/lib/Target/Hexagon/HexagonEarlyIfConv.cpp<br>
<br>
Removed: <br>
<br>
<br>
<br>
################################################################################<br>
diff --git a/llvm/lib/Target/Hexagon/HexagonEarlyIfConv.cpp
b/llvm/lib/Target/Hexagon/HexagonEarlyIfConv.cpp<br>
index b18730d967a3..d0285a7aa377 100644<br>
--- a/llvm/lib/Target/Hexagon/HexagonEarlyIfConv.cpp<br>
+++ b/llvm/lib/Target/Hexagon/HexagonEarlyIfConv.cpp<br>
@@ -282,6 +282,7 @@ bool
HexagonEarlyIfConversion::matchFlowPattern(MachineBasicBlock
*B,<br>
// can fall through into the other, in other words, it will
be executed<br>
// in both cases. We only want to predicate the block that
is executed<br>
// conditionally.<br>
+ assert(TB && FB && "Failed to find triangle
control flow blocks");<br>
unsigned TNP = TB->pred_size(), FNP =
FB->pred_size();<br>
unsigned TNS = TB->succ_size(), FNS =
FB->succ_size();<br>
<br>
<br>
<br>
<br>
_______________________________________________<br>
llvm-commits mailing list<br>
<a href="mailto:llvm-commits@lists.llvm.org" target="_blank">llvm-commits@lists.llvm.org</a><br>
<a href="https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits" rel="noreferrer" target="_blank">https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits</a><br>
</blockquote>
</div>
</blockquote>
</div>
</blockquote></div></div>