DavidSpickett wrote: If it were a GDB packet it would be in https://sourceware.org/gdb/current/onlinedocs/gdb.html/Host-I_002fO-Packets.html#Host-I_002fO-Packets but I see no sign of it, or in GDB's sources.
The first appearance of `vFile:MD5` is e0f8f574c7f41f5b61ec01aa290d52fc55a3f3c9 though it is not documented as one of our extensions in https://github.com/llvm/llvm-project/blob/main/lldb/docs/lldb-platform-packets.txt. So a good follow up PR would be to list it in that document. It's self explanatory but still, weird that it's not there. This means we could add `vFile:betterHash` (or a more general `hash:<kind>`) and try sending it to the remote. If the remote responds that it doesn't support it, we ask it for MD5. Though we may want a way to say don't use MD5 even if the remote would accept it, if security is the concern (a bit like ssh key formats work). In the case of this specific PR, it's fixing code that should have worked all along but didn't. So the issue of MD5 collision is worth looking at, but I wouldn't let it block this. @emaste, since you know about the shortcomings of MD5, could you open an issue for this? And I'll add to that the internal lldb details. https://github.com/llvm/llvm-project/pull/88812 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
