Error floating-point assist fault at ip 4000000004d08142

[LEO SIMON_1]

Hi,

I'm having error message in /var/log/messages

oracle(12479):floating-point assist fault at ip 4000000004d08142
Using rx7620 with Redhat Linux
Red Hat Enterprise Linux AS release 3 (Taroon Update 6)
Kernel 2.4.21-37.EL on an ia64
This error causing oracle intermittent down.
Do I need to upgrade the linux kernel for this issue ?

Thanks

[Mike Stroyan]

Those messages are not really errors. They only indicate that the oracle process did some math operations that required software emulation through the kernel trap handler. That happens when math is done on numbers that are very very close to zero. While the math is slower it is still done correctly, (with limited precision for such numbers).

It is possible to quiet the messages by running a program with the prctl command.
'prctl --fpemu=silent a.out'
would run a.out and any processes that it starts without reporting floating point emulation in /var/log/messages. The slow emulation will still happen. But it won't be logged.

Those messages are not related to oracle being down. That must have some other cause.

[LEO SIMON_1]

I found information from oracle
Waht about this folating-point assist fault ?

When one does computation involving floats,the result may not always be turned into normalized representation,these numbers are called "denormals" They can be thought of as really tiny numbers handles these cases,but not always does the Floating Point Unit. There are two ways to deal with this problem:
- Silently ignore it ( may be by turning the number into zero )
- Inform the user that the result is a denormal and let him do what he wants with it (=we ask the user and his software to assist FPU )
The Intel Itanium does not fully support IEEE denormals and requires software assistance to handle them.

[LEO SIMON_1]

case closed