diff mbox

[v2] ethtool: do not vzalloc(0) on registers dump

Message ID 1486038730-9469-1-git-send-email-sgruszka@redhat.com
State Accepted, archived
Delegated to: David Miller
Headers show

Commit Message

Stanislaw Gruszka Feb. 2, 2017, 12:32 p.m. UTC
If ->get_regs_len() callback return 0, we allocate 0 bytes of memory,
what print ugly warning in dmesg, which can be found further below.

This happen on mac80211 devices where ieee80211_get_regs_len() just
return 0 and driver only fills ethtool_regs structure and actually
do not provide any dump. However I assume this can happen on other
drivers i.e. when for some devices driver provide regs dump and for
others do not. Hence preventing to to print warning in ethtool code
seems to be reasonable.

ethtool: vmalloc: allocation failure: 0 bytes, mode:0x24080c2(GFP_KERNEL|__GFP_HIGHMEM|__GFP_ZERO)
<snip>
Call Trace:
[<ffffffff813bde47>] dump_stack+0x63/0x8c
[<ffffffff811b0a1f>] warn_alloc+0x13f/0x170
[<ffffffff811f0476>] __vmalloc_node_range+0x1e6/0x2c0
[<ffffffff811f0874>] vzalloc+0x54/0x60
[<ffffffff8169986c>] dev_ethtool+0xb4c/0x1b30
[<ffffffff816adbb1>] dev_ioctl+0x181/0x520
[<ffffffff816714d2>] sock_do_ioctl+0x42/0x50
<snip>
Mem-Info:
active_anon:435809 inactive_anon:173951 isolated_anon:0
 active_file:835822 inactive_file:196932 isolated_file:0
 unevictable:0 dirty:8 writeback:0 unstable:0
 slab_reclaimable:157732 slab_unreclaimable:10022
 mapped:83042 shmem:306356 pagetables:9507 bounce:0
 free:130041 free_pcp:1080 free_cma:0
Node 0 active_anon:1743236kB inactive_anon:695804kB active_file:3343288kB inactive_file:787728kB unevictable:0kB isolated(anon):0kB isolated(file):0kB mapped:332168kB dirty:32kB writeback:0kB shmem:0kB shmem_thp: 0kB shmem_pmdmapped: 0kB anon_thp: 1225424kB writeback_tmp:0kB unstable:0kB pages_scanned:0 all_unreclaimable? no
Node 0 DMA free:15900kB min:136kB low:168kB high:200kB active_anon:0kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB writepending:0kB present:15984kB managed:15900kB mlocked:0kB slab_reclaimable:0kB slab_unreclaimable:0kB kernel_stack:0kB pagetables:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB
lowmem_reserve[]: 0 3187 7643 7643
Node 0 DMA32 free:419732kB min:28124kB low:35152kB high:42180kB active_anon:541180kB inactive_anon:248988kB active_file:1466388kB inactive_file:389632kB unevictable:0kB writepending:0kB present:3370280kB managed:3290932kB mlocked:0kB slab_reclaimable:217184kB slab_unreclaimable:4180kB kernel_stack:160kB pagetables:984kB bounce:0kB free_pcp:2236kB local_pcp:660kB free_cma:0kB
lowmem_reserve[]: 0 0 4456 4456

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
---
v1 -> v2: nullify regbuf to avoid using uninitialized variable in line:
          if (regbuf && copy_to_user(useraddr, regbuf, regs.len))

 net/core/ethtool.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

Comments

John W. Linville Feb. 2, 2017, 2:27 p.m. UTC | #1
On Thu, Feb 02, 2017 at 01:32:10PM +0100, Stanislaw Gruszka wrote:
> If ->get_regs_len() callback return 0, we allocate 0 bytes of memory,
> what print ugly warning in dmesg, which can be found further below.
> 
> This happen on mac80211 devices where ieee80211_get_regs_len() just
> return 0 and driver only fills ethtool_regs structure and actually
> do not provide any dump. However I assume this can happen on other
> drivers i.e. when for some devices driver provide regs dump and for
> others do not. Hence preventing to to print warning in ethtool code
> seems to be reasonable.
> 
> ethtool: vmalloc: allocation failure: 0 bytes, mode:0x24080c2(GFP_KERNEL|__GFP_HIGHMEM|__GFP_ZERO)
> <snip>
> Call Trace:
> [<ffffffff813bde47>] dump_stack+0x63/0x8c
> [<ffffffff811b0a1f>] warn_alloc+0x13f/0x170
> [<ffffffff811f0476>] __vmalloc_node_range+0x1e6/0x2c0
> [<ffffffff811f0874>] vzalloc+0x54/0x60
> [<ffffffff8169986c>] dev_ethtool+0xb4c/0x1b30
> [<ffffffff816adbb1>] dev_ioctl+0x181/0x520
> [<ffffffff816714d2>] sock_do_ioctl+0x42/0x50
> <snip>
> Mem-Info:
> active_anon:435809 inactive_anon:173951 isolated_anon:0
>  active_file:835822 inactive_file:196932 isolated_file:0
>  unevictable:0 dirty:8 writeback:0 unstable:0
>  slab_reclaimable:157732 slab_unreclaimable:10022
>  mapped:83042 shmem:306356 pagetables:9507 bounce:0
>  free:130041 free_pcp:1080 free_cma:0
> Node 0 active_anon:1743236kB inactive_anon:695804kB active_file:3343288kB inactive_file:787728kB unevictable:0kB isolated(anon):0kB isolated(file):0kB mapped:332168kB dirty:32kB writeback:0kB shmem:0kB shmem_thp: 0kB shmem_pmdmapped: 0kB anon_thp: 1225424kB writeback_tmp:0kB unstable:0kB pages_scanned:0 all_unreclaimable? no
> Node 0 DMA free:15900kB min:136kB low:168kB high:200kB active_anon:0kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB writepending:0kB present:15984kB managed:15900kB mlocked:0kB slab_reclaimable:0kB slab_unreclaimable:0kB kernel_stack:0kB pagetables:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB
> lowmem_reserve[]: 0 3187 7643 7643
> Node 0 DMA32 free:419732kB min:28124kB low:35152kB high:42180kB active_anon:541180kB inactive_anon:248988kB active_file:1466388kB inactive_file:389632kB unevictable:0kB writepending:0kB present:3370280kB managed:3290932kB mlocked:0kB slab_reclaimable:217184kB slab_unreclaimable:4180kB kernel_stack:160kB pagetables:984kB bounce:0kB free_pcp:2236kB local_pcp:660kB free_cma:0kB
> lowmem_reserve[]: 0 0 4456 4456
> 
> Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
> ---
> v1 -> v2: nullify regbuf to avoid using uninitialized variable in line:
>           if (regbuf && copy_to_user(useraddr, regbuf, regs.len))
> 
>  net/core/ethtool.c | 9 ++++++---
>  1 file changed, 6 insertions(+), 3 deletions(-)
> 
> diff --git a/net/core/ethtool.c b/net/core/ethtool.c
> index 6b3eee0..4b6dc9b 100644
> --- a/net/core/ethtool.c
> +++ b/net/core/ethtool.c
> @@ -1405,9 +1405,12 @@ static int ethtool_get_regs(struct net_device *dev, char __user *useraddr)
>  	if (regs.len > reglen)
>  		regs.len = reglen;
>  
> -	regbuf = vzalloc(reglen);
> -	if (reglen && !regbuf)
> -		return -ENOMEM;
> +	regbuf = NULL;

Any reason to prefer this over changing the declaration to include
the assignment?

	void *regbuf = NULL;

> +	if (reglen) {
> +		regbuf = vzalloc(reglen);
> +		if (!regbuf)
> +			return -ENOMEM;
> +	}
>  
>  	ops->get_regs(dev, &regs, regbuf);
>  
> -- 
> 1.8.3.1
> 
>
Stanislaw Gruszka Feb. 2, 2017, 3:03 p.m. UTC | #2
On Thu, Feb 02, 2017 at 09:27:18AM -0500, John W. Linville wrote:
> > -	regbuf = vzalloc(reglen);
> > -	if (reglen && !regbuf)
> > -		return -ENOMEM;
> > +	regbuf = NULL;
> 
> Any reason to prefer this over changing the declaration to include
> the assignment?
> 
> 	void *regbuf = NULL;

I've chosen this form to have initialization near the vzalloc() call,
after sanity checks, however I don't think it's better or worse over
declaration initialization.

Stanislaw
David Miller Feb. 3, 2017, 4:11 p.m. UTC | #3
From: Stanislaw Gruszka <sgruszka@redhat.com>
Date: Thu,  2 Feb 2017 13:32:10 +0100

> If ->get_regs_len() callback return 0, we allocate 0 bytes of memory,
> what print ugly warning in dmesg, which can be found further below.
> 
> This happen on mac80211 devices where ieee80211_get_regs_len() just
> return 0 and driver only fills ethtool_regs structure and actually
> do not provide any dump. However I assume this can happen on other
> drivers i.e. when for some devices driver provide regs dump and for
> others do not. Hence preventing to to print warning in ethtool code
> seems to be reasonable.
 ...
> Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
> ---
> v1 -> v2: nullify regbuf to avoid using uninitialized variable in line:
>           if (regbuf && copy_to_user(useraddr, regbuf, regs.len))

Applied, thanks.
diff mbox

Patch

diff --git a/net/core/ethtool.c b/net/core/ethtool.c
index 6b3eee0..4b6dc9b 100644
--- a/net/core/ethtool.c
+++ b/net/core/ethtool.c
@@ -1405,9 +1405,12 @@  static int ethtool_get_regs(struct net_device *dev, char __user *useraddr)
 	if (regs.len > reglen)
 		regs.len = reglen;
 
-	regbuf = vzalloc(reglen);
-	if (reglen && !regbuf)
-		return -ENOMEM;
+	regbuf = NULL;
+	if (reglen) {
+		regbuf = vzalloc(reglen);
+		if (!regbuf)
+			return -ENOMEM;
+	}
 
 	ops->get_regs(dev, &regs, regbuf);