Re: broken JIT support on Fedora 40
Thomas Munro <thomas.munro@gmail.com>
From: Thomas Munro <thomas.munro@gmail.com>
To: Dmitry Dolgov <9erthalion6@gmail.com>
Cc: Daniel Gustafsson <daniel@yesql.se>,
Pavel Stehule <pavel.stehule@gmail.com>, Robert Haas <robertmhaas@gmail.com>,
PostgreSQL Hackers <pgsql-hackers@postgresql.org>
Date: 2024-03-30T23:49:36Z
Lists: pgsql-hackers
On Sun, Mar 31, 2024 at 5:59 AM Dmitry Dolgov <9erthalion6@gmail.com> wrote: > Yeah, sorry, I'm a bit baffled about this situation myself. Yesterday > I've opened a one-line PR fix that should address the issue, maybe this > would help. In the meantime I've attached what did work for me as a > workaround -- it essentially just makes the deform function to return > some value. It's ugly, but since call site will ignore that, and it's > only one occasion where LLVMBuildRetVoid is used, maybe it's acceptable. > Give me a moment, I'm going to test this change more (waiting on > rebuilding LLVM, it takes quire a while on my machine :( ), then can > confirm that it works as expected on the latest version. Great news. I see your PR: https://github.com/llvm/llvm-project/pull/87093 Checking their release schedule, they have: Mar 5th: 18.1.0 was released Mar 8th: 18.1.1 was released Mar 19th: 18.1.2 was released Apr 16th: 18.1.3 Apr 30th: 18.1.4 May 14th: 18.1.5 May 28th: 18.1.6 (if necessary) Our next release is May 9th. So assuming your PR goes in in the next couple of weeks and makes it into their 18.1.3 or even .4, there is no point in pushing a work-around on our side.
Commits
-
Run LLVM verify pass on IR in assert builds.
- 8583b1f99340 18.0 landed
-
Fix illegal attribute propagation in LLVM JIT.
- 01b55203ace0 12.19 landed
- 4f90750b5326 13.15 landed
- 7fe32eaa4a87 14.12 landed
- 74992929a79e 15.7 landed
- 53c8d6c9f157 17.0 landed
- bf1cfe77ed49 16.3 landed
-
Track LLVM 18 changes.
- d282e88e5052 17.0 cited
-
jit: Changes for LLVM 17.
- 76200e5ee469 17.0 cited