[gentoo-embedded] can't resolve symbol '__divsi3' on armv5tel-softfloat-linux-uclibc

[gentoo-embedded] can't resolve symbol '__divsi3' on armv5tel-softfloat-linux-uclibc

[KiberGus]

[-- Attachment #1: Type: text/plain, Size: 481 bytes --]

I'm trying to install gentoo on my PDA. And I have found that some programms
(such as nano, python, nm) fail with srting like this:
nano: can't resolve symbol '__divsi3'
python: can't resolve symbol '__udivsi3'

I can't underrstand, how to fix this. Math libraries work OK, my simple test
application, which uses floating point operation work fine. Does any body
know, what operation __udivsi3 performs? It is clearly, that it is divide,
but what type it's arguments have?
Thanks.

[-- Attachment #2: Type: text/html, Size: 517 bytes --]

Re: [gentoo-embedded] can't resolve symbol '__divsi3' on armv5tel-softfloat-linux-uclibc

[Derick Swanepoel]

On 25/07/06, KiberGus <kibergus@gmail.com> wrote:
> I'm trying to install gentoo on my PDA. And I have found that some programms
> (such as nano, python, nm) fail with srting like this:
> nano: can't resolve symbol '__divsi3'
> python: can't resolve symbol '__udivsi3'
>
>  I can't underrstand, how to fix this. Math libraries work OK, my simple
> test application, which uses floating point operation work fine. Does any
> body know, what operation __udivsi3 performs? It is clearly, that it is
> divide, but what type it's arguments have?
> Thanks.

I have the same problem on armeb-softfloat-linux-uclibc. I also have
apps that fail with __fixdfsi. Many of the busybox applets fail with
__udivsi3 (e.g. awk, ping, top, df, netstat, fdisk) - basically
anything that does a certain kind of floating point divide. They only
fail when that specific instruction is reached, so for example ash
would only fail if mod or div is used e.g. "echo $((5 % 2))" or "echo
$((16 / 4))".

So far I have been unable to construct a test case - my own test app
that does floating point div also works.

I suspect that it is a bug in gcc and the way it generates code on ARM
softfloat targets, but I have no idea how to find out.

Derick
-- 
gentoo-embedded@gentoo.org mailing list

Re: [gentoo-embedded] can't resolve symbol '__divsi3' on armv5tel-softfloat-linux-uclibc

[Anish Patel]

Derick Swanepoel wrote:
> On 25/07/06, KiberGus <kibergus@gmail.com> wrote:
>> I'm trying to install gentoo on my PDA. And I have found that some 
>> programms
>> (such as nano, python, nm) fail with srting like this:
>> nano: can't resolve symbol '__divsi3'
>> python: can't resolve symbol '__udivsi3'
>>
>>  I can't underrstand, how to fix this. Math libraries work OK, my simple
>> test application, which uses floating point operation work fine. Does any
>> body know, what operation __udivsi3 performs? It is clearly, that it is
>> divide, but what type it's arguments have?
>> Thanks.
> 
> I have the same problem on armeb-softfloat-linux-uclibc. I also have
> apps that fail with __fixdfsi. Many of the busybox applets fail with
> __udivsi3 (e.g. awk, ping, top, df, netstat, fdisk) - basically
> anything that does a certain kind of floating point divide. They only
> fail when that specific instruction is reached, so for example ash
> would only fail if mod or div is used e.g. "echo $((5 % 2))" or "echo
> $((16 / 4))".
> 
> So far I have been unable to construct a test case - my own test app
> that does floating point div also works.
> 
> I suspect that it is a bug in gcc and the way it generates code on ARM
> softfloat targets, but I have no idea how to find out.
> 
> Derick
i have also had the same problem with some Intel CSR stuff and the newer 
  kernels, with 2.6.12 i don't have problems armeb-unknown-linux-gnu or 
anything else.

Anish

-- 
gentoo-embedded@gentoo.org mailing list

Re: [gentoo-embedded] can't resolve symbol '__divsi3' on armv5tel-softfloat-linux-uclibc

[Mike Frysinger]

[-- Attachment #1: Type: text/plain, Size: 394 bytes --]

On Tuesday 25 July 2006 14:16, KiberGus wrote:
> I'm trying to install gentoo on my PDA. And I have found that some
> programms (such as nano, python, nm) fail with srting like this:
> nano: can't resolve symbol '__divsi3'
> python: can't resolve symbol '__udivsi3'

that symbol should be provided by libgcc_s.so.1 which comes from your 
toolchain and should be located in /lib/
-mike

[-- Attachment #2: Type: application/pgp-signature, Size: 827 bytes --]