<p>it&#39;s apparently fully documented at <a href="https://sourceware.org/gdb/current/onlinedocs/gdb.html/Agent-Expressions.html" target="_blank" rel="nofollow noopener" translate="no"><span class="invisible">https://</span><span class="ellipsis">sourceware.org/gdb/current/onl</span><span class="invisible">inedocs/gdb.html/Agent-Expressions.html</span></a></p>

<p><span class="h-card" translate="no"><a href="https://mastodon.social/@whitequark" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>whitequark</span></a></span> sadly yes</p>

<p><span class="h-card" translate="no"><a href="https://glauca.space/@r" class="u-url mention">@<span>r</span></a></span> ax.def</p>

<p>it&#39;s a stack machine that operates on integer values with all the usual suspects (const32, goto, bit_or, dup, pop, rot), but... but! like x86, it has prefixes</p><p>specifically, the `float` prefix modifies the next operation to be floating-point instead of integer. like world&#39;s worst x86 variant</p>

<p>did you know that GDB includes a bytecode compiler? specifically, it has a private bytecode format used between it and the GDB server, which the latter uses to implement conditional breakpoints with complex expressions and tracepoints</p><p>every time it hits a breakpoint/tracepoint it evaluates the bytecode, which has jumps and can read arbitrary target memory, and decides whether it really was hit or not</p>

<p><span class="h-card" translate="no"><a href="https://mastodon.social/@whitequark" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>whitequark</span></a></span> the llvm stack is chock-full of assertions used to validate inputs/usage... I hate it so much</p>

<p>filed <a href="https://github.com/llvm/llvm-project/issues/138536" target="_blank" rel="nofollow noopener" translate="no"><span class="invisible">https://</span><span class="ellipsis">github.com/llvm/llvm-project/i</span><span class="invisible">ssues/138536</span></a></p>

<p><span class="h-card" translate="no"><a href="https://mastodon.social/@whitequark" class="u-url mention">@<span>whitequark</span></a></span> yeah I think you might be right. I think I was converting it incorrectly somehow</p>

<p><span class="h-card" translate="no"><a href="https://digipres.club/@foone" class="u-url mention">@<span>foone</span></a></span> that sounds wrong to me?</p>