[committed] [PR99454] LRA: Process 0..9 constraints in process_address_1

Vladimir Makarov vmakarov@redhat.com
Tue Mar 9 15:26:58 GMT 2021


On 2021-03-09 9:53 a.m., Martin Liška wrote:
> On 3/9/21 3:12 PM, Vladimir Makarov via Gcc-patches wrote:
>> The patch was successfully bootstrapped and tested on x86-64, 
>> ppc64le, and arm64.  Unfortunately, I did not manage to reduce the 
>> test (whose size is 5MB).
>
> I've just reduced test-case from it:
>
> $ cat a.c
> struct skb_shared_info {
>   short gso_size;
> };
>
> enum { NETDEV_TX_OK };
>
> struct iphdr {
>   short tot_len;
>   int daddr;
> };
>
> int tg3_tso_bug();
> int netdev_priv();
> int skb_cow_head();
> int tcp_hdrlen__builtin_expect();
> struct iphdr *ip_hdr();
> int _tg3_flag();
> int tg3_tso_bug_gso_check();
>
> int
> tg3_start_xmit() {
>   int *tp = netdev_priv();
>   int mss, tnapi;
>   struct iphdr *iph;
>   tnapi = mss = ((struct skb_shared_info *)0)->gso_size;
>   if (mss) {
>     int hdr_len;
>     if (skb_cow_head())
>       iph = ip_hdr();
>     hdr_len = tcp_hdrlen__builtin_expect() && _tg3_flag();
>     if (tg3_tso_bug_gso_check())
>       return tg3_tso_bug(tp, tnapi);
>     iph->tot_len = mss + hdr_len;
>     if (_tg3_flag(tp) || tp)
>       ;
>     else
>       asm("" : : "g"(iph->daddr));
>   }
>   return 0;
> }
>
> Please add it into test-suite.
>
Thank you.  Will do.




More information about the Gcc-patches mailing list