Tiny Core Linux

Tiny Core Base => TCB Bugs => Topic started by: dentonlt on March 28, 2014, 08:36:12 PM

Title: 5.2: busybox dc erroneous output
Post by: dentonlt on March 28, 2014, 08:36:12 PM
busybox dc runs, but its output is always the same (perhaps an error code value or just MAX): 6.05096e-270.

Try: 'dc 10 1 add p' or 'dc 10 100 div p'

Since this occurs in base, I expect it's going to be a problem until 5.3 at least the next 5.3rc.

Same problem on both 32-bit and Pure 64. -

DLT
Title: Re: 5.2: busybox dc erroneous output
Post by: bmarkus on March 29, 2014, 01:05:39 AM
I see the same in the x86. It works properly in piCore 5.2 on the Raspberry Pi.
Title: Re: 5.2: busybox dc erroneous output
Post by: curaga on March 30, 2014, 05:04:23 AM
It still happens in 1.22.1 too. Will check further.
Title: Re: 5.2: busybox dc erroneous output
Post by: Juanito on March 30, 2014, 05:10:36 AM
see above, it works with x86_64, but not x86...
Title: Re: 5.2: busybox dc erroneous output
Post by: bmarkus on March 30, 2014, 05:38:22 AM
see above, it works with x86_64, but not x86...

Didn't check the new one, but in my previous test build in 4.7.7 against eglibc an using the older GCC it is working.
Title: Re: 5.2: busybox dc erroneous output
Post by: curaga on March 31, 2014, 04:35:14 AM
It turned out to be a GCC bug. I propose a rebuild of the current version on 4.x, to minimize any possible issues from a version upgrade at the same time.
Title: Re: 5.2: busybox dc erroneous output
Post by: Juanito on March 31, 2014, 05:18:17 AM
..strange that gcc-4.7.2 is used for both core and corepure64, but dc only fails with x86
Title: Re: 5.2: busybox dc erroneous output
Post by: curaga on March 31, 2014, 05:53:34 AM
It's some combination of optimization that triggers it, so not seeing on another arch is expected.
Title: Re: 5.2: busybox dc erroneous output
Post by: bmarkus on March 31, 2014, 05:54:50 AM
Which GCC versions have this bug?
Title: Re: 5.2: busybox dc erroneous output
Post by: curaga on March 31, 2014, 11:27:47 AM
I don't know exactly, beyond 4.7.2.