I have some very, very old (ancient, in fact) 32 bit statically linked executables run as part of a testbed.
With the new 5.1.6-300.fc30.x86_64 kernel installed this morning, they are all failing because the programs are confused by something in the initial code before they even get to the main program.
Anyone know if there is some major new difference in i386 support in the new kernel?
(When I get a chance I'll boot the old kernel and make sure it still works and this isn't something different than I suspect).
On Thu, 6 Jun 2019 08:19:17 -0400 Tom Horsley wrote:
(When I get a chance I'll boot the old kernel and make sure it still works and this isn't something different than I suspect).
Yep, the tests run correctly with the 5.0.17 kernel booted. The programs were only linked 16 years ago :-).
On Thu, 2019-06-06 at 09:56 -0400, Tom Horsley wrote:
On Thu, 6 Jun 2019 08:19:17 -0400 Tom Horsley wrote:
(When I get a chance I'll boot the old kernel and make sure it still works and this isn't something different than I suspect).
Yep, the tests run correctly with the 5.0.17 kernel booted. The programs were only linked 16 years ago :-
Are this files in a.out format? 5.1 removes support for a.out: https://www.phoronix.com/scan.php?page=news_item&px=Linux-Dropping-A.Out
On Thu, 06 Jun 2019 18:53:01 +0200 Louis Lagendijk wrote:
Are this files in a.out format? 5.1 removes support for a.out: https://www.phoronix.com/scan.php?page=news_item&px=Linux-Dropping-A.Out
Nah. I think they are calling a system service that didn't exist (which they were prepared to ignore), but now it does exist and doesn't do anything remotely like the one it thought it was calling (which existed in a custom kernel 16 years ago :-).
By the way, I notice that strace still doesn't recognize the new services even though they have made it into the kernel now.
-
Louis Lagendijk -
Tom Horsley