Re: [fuse-devel] copy_out_args() returning -EINVAL

[Paul N. <pa...@ps...>]

So I think I've solved my problem.  It seems that I was unclear on the 
fuse_reply_* api.  I was using fuse_reply_entry() instead of 
fuse_reply_create()!  I'm still new to the low level routines so I'm not 
surprised I made this mistake.  However, debugging would have been 
easier had fuse not accepted a reply from what turned out to be the 
incorrect function.
paul

Paul Nowoczynski wrote:
> Hi,
> I'm working on a low-level interface and I've come across a problem with 
> create() ops where my app sees an EIO.  The filesystem backend appears 
> to perform the correct operations and the fuse lib reports a 0 rc before 
> reporting to the kernel.
>
> After confirming that I was returning the correct inode number and mode, 
> I did some digging and discovered that the problem is in kernel/dev.c.  
> The copy_out_args() call is returning -EINVAL because this test is failing:
>
>         if (reqsize < nbytes || (reqsize > nbytes && !out->argvar))
>                 return -EINVAL;
>
> Using printk I obtained the values for reqsize and friends:
>     reqsize=152 nbytes=136 out->argvar=0
>
> I have a feeling that this error is being caused by an inconsistency 
> between the fuse kernel module and the library or headers.  But I've 
> verified that kernel module, libs, and headers are all based on the same 
> version (2.7.2). 
>
> Does anyone have a good suggestion?  I'm running a suse kernel - 
> 2.6.16-27-0.9 and I doubt that upgrading it is an option..
> thanks,
> paul
>
> ------------------------------------------------------------------------------
> This SF.net email is sponsored by:
> SourcForge Community
> SourceForge wants to tell your story.
> http://p.sf.net/sfu/sf-spreadtheword
> _______________________________________________
> fuse-devel mailing list
> fus...@li...
> https://lists.sourceforge.net/lists/listinfo/fuse-devel
>