forttl severe (24)

[Guglielmo_Monaco]

Dear Sirs,
I am running an intel compiled fortran program which writes and reads very long fort.4 files (5G up to 250 G).
For file smaller than 20G everything is OK, but for larger files, I have errors of the type

forrtl: severe (24): end-of-file during read, unit -4, file stdin
Image PC Routine Line Source
mo400 00000000004BA34D Unknown Unknown Unknown
mo400 00000000004B8E55 Unknown Unknown Unknown
mo400 0000000000466330 Unknown Unknown Unknown
mo400 000000000042B51F Unknown Unknown Unknown
mo400 000000000042AD52 Unknown Unknown Unknown
mo400 000000000044306C Unknown Unknown Unknown
28,1 2%
mo400 000000000042AD52 Unknown Unknown Unknown
mo400 000000000044306C Unknown Unknown Unknown
mo400 000000000040BD34 Unknown Unknown Unknown
mo400 0000000000404FC5 Unknown Unknown Unknown
mo400 00000000004038AE Unknown Unknown Unknown
mo400 00000000004032EC Unknown Unknown Unknown
libc.so.6 000000397421EB1D Unknown Unknown Unknown
mo400 00000000004031E9 Unknown Unknown Unknown

My machine has the following uname -a
Linux *.*.* 2.6.31.5-127.fc12.x86_64 #1 SMP Sat Nov 7 21:11:14 EST 2009 x86_64 x86_64 x86_64 GNU/Linux

I have read on your website of problems deriving fromn reading/writing with different compilers, but this is not my case.
Any suggestion is appreciated.
Best regards
GM

[mecej4]

The indications are that your program is trying to read more data than is in the file.

In these circumstances, aborting the program is a normal response, and it is up to you to find out why your program attempts to read beyond end-of-file.

You may find it easier to pinpoint the reason if you enable traceback or if you add diagnostic WRITE statements to your program at appropriate places.

I find it a little odd that Unit-4 is connected to stdin, but that may have good reasons behind it.

[Ron_Green]

also, was the file.4 created by Intel Fortran or is it from another system?

As mentioned above, compile with -O0 -g -traceback and then examine the READ statement that is failing. You may want to put in a IOSTAT or the newer IOMSG= specifier on the read along with ERR= or END= stat

it does appear that the file is truncated or your read is just going too far.

ron