Re: [Xen-devel] bogus wrap check in xen-netback

[Juergen Gross <jgross@xxxxxxxx>]

On 25/04/18 09:19, Olaf Hering wrote:
> With commit 48856286b64e ("xen/netback: shutdown the ring if it contains 
> garbage.") a new check was added to xen-netback, which triggers for me.
> 
> Since there are bugs in ring buffer handling I tried to trigger them earlier 
> by changing RING_IDX from u32 to u16. Now I found another one, and I wonder 
> if the error below could potentially also hit with u32:
> 
> ...
> [  624.186492] br0: port 3(vif2.0) entered forwarding state
> [  624.186522] br0: port 3(vif2.0) entered forwarding state
> [  680.865398] vif vif-1-0 vif1.0: Impossible number of requests. req_prod 0, 
> req_cons 65400, size 256
> [  680.865402] vif vif-1-0 vif1.0: fatal error; disabling device
> [  680.865495] br0: port 2(vif1.0) entered disabled state
> [  689.433849] vif vif-2-0 vif2.0: Impossible number of requests. req_prod 0, 
> req_cons 65527, size 256
> [  689.433857] vif vif-2-0 vif2.0: fatal error; disabling device
> [  689.433945] br0: port 3(vif2.0) entered disabled state
> [  690.930512] pktgen: Packet Generator for packet performance testing. 
> Version: 2.75
> ...
> 
> What exactly is that check in xenvif_tx_build_gops trying to achieve? 
> Subtracting a non-zero value from zero will always create something larger 
> than XEN_NETIF_TX_RING_SIZE.

Why? (u16)0 - (u16)65400 == 136


Juergen

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel