Re: select(0, ..) is valid ?

[Steve Fox <drfickle@linux.vnet.ibm.com>]

On Wed, 2007-05-16 at 17:59 -0700, Badari Pulavarty wrote:

> Here it is ..
> 
> Should I do one for poll() also ?
> 
> Thanks,
> Badari
> 
> Optimize select by a using stack space for small fd sets.
> core_sys_select() already has this optimization. This is
> for compat version. 
> 
> Signed-off-by: Badari Pulavarty <pbadari@us.ibm.com>
> ---
>  fs/compat.c |   17 +++++++++++------
>  1 file changed, 11 insertions(+), 6 deletions(-)
> 
> Index: linux-2.6.22-rc1/fs/compat.c
> ===================================================================
> --- linux-2.6.22-rc1.orig/fs/compat.c	2007-05-12 18:45:56.000000000 -0700
> +++ linux-2.6.22-rc1/fs/compat.c	2007-05-16 17:50:39.000000000 -0700
> @@ -1544,9 +1544,10 @@ int compat_core_sys_select(int n, compat
>  	compat_ulong_t __user *outp, compat_ulong_t __user *exp, s64 *timeout)
>  {
>  	fd_set_bits fds;
> -	char *bits;
> +	void *bits;
>  	int size, max_fds, ret = -EINVAL;
>  	struct fdtable *fdt;
> +	long stack_fds[SELECT_STACK_ALLOC/sizeof(long)];
> 
>  	if (n < 0)
>  		goto out_nofds;
> @@ -1564,11 +1565,14 @@ int compat_core_sys_select(int n, compat
>  	 * since we used fdset we need to allocate memory in units of
>  	 * long-words.
>  	 */
> -	ret = -ENOMEM;
>  	size = FDS_BYTES(n);
> -	bits = kmalloc(6 * size, GFP_KERNEL);
> -	if (!bits)
> -		goto out_nofds;
> +	bits = stack_fds;
> +	if (size > sizeof(stack_fds) / 6) {
> +		bits = kmalloc(6 * size, GFP_KERNEL);
> +		ret = -ENOMEM;
> +		if (!bits)
> +			goto out_nofds;
> +	}
>  	fds.in      = (unsigned long *)  bits;
>  	fds.out     = (unsigned long *) (bits +   size);
>  	fds.ex      = (unsigned long *) (bits + 2*size);
> @@ -1600,7 +1604,8 @@ int compat_core_sys_select(int n, compat
>  	    compat_set_fd_set(n, exp, fds.res_ex))
>  		ret = -EFAULT;
>  out:
> -	kfree(bits);
> +	if (bits != stack_fds)
> +		kfree(bits);
>  out_nofds:
>  	return ret;
>  }

Andy put this through a couple machines on test.kernel.org and elm3b6
was fixed, however elm3b239 still had a boot error.

BUG: at mm/slab.c:777 __find_general_cachep()

Call Trace:
 [<ffffffff802729c6>] __kmalloc+0xa6/0xe0
 [<ffffffff8021d21b>] cache_k8_northbridges+0x9b/0x120
 [<ffffffff80688af3>] gart_iommu_init+0x33/0x5b0
 [<ffffffff802211a3>] __wake_up+0x43/0x70
 [<ffffffff80453b90>] genl_rcv+0x0/0x70
 [<ffffffff80452175>] netlink_kernel_create+0x155/0x170
 [<ffffffff80684029>] pci_iommu_init+0x9/0x20
 [<ffffffff8067e6f4>] kernel_init+0x154/0x330
 [<ffffffff8020a8d8>] child_rip+0xa/0x12
 [<ffffffff80348e10>] acpi_ds_init_one_object+0x0/0x7c
 [<ffffffff8067e5a0>] kernel_init+0x0/0x330
 [<ffffffff8020a8ce>] child_rip+0x0/0x12

See the "2.6.22-rc2-git1 +1" row at http://test.kernel.org/ for full
logs.

-- 

Steve Fox
IBM Linux Technology Center