Re: [PATCH][next] net/ipv4: Replace one-element array with flexible-array member

2 messages, 1 author, 2021-08-02 · open the first message on its own page

Re: [PATCH][next] net/ipv4: Replace one-element array with flexible-array member

From: Gustavo A. R. Silva <hidden>
Date: 2021-08-02 18:43:56


On 8/2/21 09:20, patchwork-bot+netdevbpf@kernel.org wrote:
Hello:

This patch was applied to netdev/net-next.git (refs/heads/master):

On Sat, 31 Jul 2021 12:08:30 -0500 you wrote:
quoted
There is a regular need in the kernel to provide a way to declare having
a dynamically sized set of trailing elements in a structure. Kernel code
should always use “flexible array members”[1] for these cases. The older
style of one-element or zero-length arrays should no longer be used[2].

Use an anonymous union with a couple of anonymous structs in order to
keep userspace unchanged:

[...]
Here is the summary with links:
  - [next] net/ipv4: Replace one-element array with flexible-array member
    https://git.kernel.org/netdev/net-next/c/2d3e5caf96b9
arghh... this has a bug. Sorry, Dave. I will send a fix for this, shortly.

--
Gustavo

Re: [PATCH][next] net/ipv4: Replace one-element array with flexible-array member

From: Gustavo A. R. Silva <hidden>
Date: 2021-08-02 19:15:25


On 8/2/21 13:46, Gustavo A. R. Silva wrote:

On 8/2/21 09:20, patchwork-bot+netdevbpf@kernel.org wrote:
quoted
Hello:

This patch was applied to netdev/net-next.git (refs/heads/master):

On Sat, 31 Jul 2021 12:08:30 -0500 you wrote:
quoted
There is a regular need in the kernel to provide a way to declare having
a dynamically sized set of trailing elements in a structure. Kernel code
should always use “flexible array members”[1] for these cases. The older
style of one-element or zero-length arrays should no longer be used[2].

Use an anonymous union with a couple of anonymous structs in order to
keep userspace unchanged:

[...]
Here is the summary with links:
  - [next] net/ipv4: Replace one-element array with flexible-array member
    https://git.kernel.org/netdev/net-next/c/2d3e5caf96b9
arghh... this has a bug. Sorry, Dave. I will send a fix for this, shortly.
BTW... can we expect msf->imsf_numsrc to be zero at some point in the following
pieces of code?

net/ipv4/igmp.c:
2553         copycount = count < msf->imsf_numsrc ? count : msf->imsf_numsrc;
2554         len = copycount * sizeof(psl->sl_addr[0]);
2555         msf->imsf_numsrc = count;
2556         if (put_user(IP_MSFILTER_SIZE(copycount), optlen) ||
2557             copy_to_user(optval, msf, IP_MSFILTER_SIZE(0))) {
2558                 return -EFAULT;
2559         }

net/ipv4/ip_sockglue.c:
1228         case IP_MSFILTER:
1229         {
1230                 struct ip_msfilter *msf;
1231
1232                 if (optlen < IP_MSFILTER_SIZE(0))
1233                         goto e_inval;
1234                 if (optlen > sysctl_optmem_max) {
1235                         err = -ENOBUFS;
1236                         break;
1237                 }
1238                 msf = memdup_sockptr(optval, optlen);
1239                 if (IS_ERR(msf)) {

			...

1250                 if (IP_MSFILTER_SIZE(msf->imsf_numsrc) > optlen) {
1251                         kfree(msf);
1252                         err = -EINVAL;
1253                         break;
1254                 }
1255                 err = ip_mc_msfilter(sk, msf, 0);
1256                 kfree(msf);
1257                 break;
1258         }

Thanks
--
Gustavo
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help