<div dir="ltr"><div dir="auto"><div><div class="gmail_quote"><div dir="ltr">On Tue, 28 Aug 2018, 17:14 Artem Dergachev via cfe-dev, <<a href="mailto:cfe-dev@lists.llvm.org" target="_blank">cfe-dev@lists.llvm.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">In various debug dumps (eg., Clang's -ast-dump), various objects (eg., <br>
Stmts and Decls in that -ast-dump) are identified by pointers. It's very <br>
reliable in the sense that no two objects would ever have the same <br>
pointer at the same time, but it's unpleasant that pointers change <br>
across runs. Having deterministic identifiers instead of pointers would <br>
aid debugging: imagine a conditional break by object identifier that has <br>
not yet been constructed, or simply trying to align two debug dumps of <br>
different kind from different runs together. Additionally, pointers are <br>
hard to read and memorize; it's hard to notice the difference between <br>
0x7f80a28325e0 and 0x7f80a28325a0, especially when they're a few screens <br>
apart.<br>
<br>
Hence the idea: why don't we print the offset into the allocator's <br>
memory slab instead of a pointer?</blockquote></div></div><div dir="auto"><br></div><div dir="auto">Make this "as well as" rather than "instead of" and it sounds great to me. When debugging, it's useful to be able to dump a large complex object, find the piece you want, grab its address and start accessing it directly.</div><div dir="auto"><br></div><div>(For the pointer stability problem, at least on Linux you can turn off ASLR. When running under gdb, that's typically done for you, and you can do it manually with setarch. But it would be nice to have an easier way to identify objects than a long, essentially meaningless address.)</div><div dir="auto"><br></div><div dir="auto"><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">We use BumpPtrAllocator all over the <br>
place, which boils down to a set of slabs on which all objects are <br>
placed in the order in which they are allocated. It is easy for the <br>
allocator to identify if a pointer belongs to that allocator, and if so, <br>
deteremine which slab it belongs to and at what offset the object is in <br>
that slab. Therefore it is possible to identify the object by its (slab <br>
index, offset) pair. Eg., "TypedefDecl 0:528" (you already memorized it) <br>
instead of "TypedefDecl 0x7f80a28325e0". This could be applied to all <br>
sorts of objects that live in BumpPtrAllocators.<br>
<br>
In order to compute such identifier, we only need access to the object <br>
and to the allocator. No additional memory is used to store such <br>
identifier. Such identifier would also be persistent across runs as long <br>
as the same objects are allocated in the same order, which is, i <br>
suspect, often the case.<br>
<br>
One of the downsides of this identifier is that it's not going to be the <br>
same on different machines, because the same data structure may require <br>
different amounts of memory on different hosts. So it wouldn't <br>
necessarily help understanding a dump that the user sent you. But it <br>
still seems to be better than pointers.<br>
<br>
Should we go ahead and try to implement it?<br>
_______________________________________________<br>
cfe-dev mailing list<br>
<a href="mailto:cfe-dev@lists.llvm.org" rel="noreferrer" target="_blank">cfe-dev@lists.llvm.org</a><br>
<a href="http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-dev" rel="noreferrer noreferrer" target="_blank">http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-dev</a><br>
</blockquote></div></div></div>
</div>