On 10/29/07, michael.firth@bt.com michael.firth@bt.com wrote:
-----Original Message----- From: Rune Torgersen [mailto:runet@innovsys.com] Sent: 29 October 2007 21:35 To: Firth,MJC,Michael,DMM R; timur@freescale.com; grant.likely@secretlab.ca Cc: u-boot-users@lists.sourceforge.net; gerald.vanbaren@ge.com Subject: RE: [U-Boot-Users] Bug in malloc()?
From: michael.firth@bt.com
-----Original Message----- From: Rune Torgersen [mailto:runet@innovsys.com]
From: Timur Tabi [mailto:timur@freescale.com] Sent: Monday, October 29, 2007 3:55 PM
So *all* versions of gcc 4.1 are broken, or just newer ones? Do we know exactly which versions are broken?
Lerts start a list.... Broken: gcc 3.4.3, glibc 2.3.4, binutils 2.15.94 gcc 4.1.2, glibc 2.5, binutils 2.17
From: glikely@secretlab.ca [mailto:glikely@secretlab.ca]
On Behalf
Of Grant Likely
Actually, #3 is migrate to an *older* compiler version. :-(
Unfortunately EABI relocation seems to be a rather
undocumented thing.
So, from Rune's list, it seems you actually need a
'middling' compiler
version, too old won't work, and too new won't work!
Sadly, I'm currently using gcc-3.4.3 for my U-Boot builds, so I'll probably try one of the other two fixes, at least as an interim measure, to prove that otherwise my u-boot-1.3.0 build is working.
Then use solution #1. (works on both 3.4.3 amd 4.1.2 for me)
Will try that, it also has the advantage of being the easiest of the 3 options to try. Are there any bad side effects of disabling that define, given that Grant's not keen to back it out more generally?
The badness is that it's just avoiding the problem. The *real* problem is that our C environment is insane and pointers to static data need to be manually adjusted at runtime.
g.