Re: Missing NULL check

[Fangrui Song <i@maskray.me>]

On Tue, Sep 12, 2023 at 11:50 AM Torbjorn SVENSSON via Binutils
<binutils@sourceware.org> wrote:
>
>
>
> On 2023-09-12 17:40, jacob navia wrote:
> > This is very similar to the last one:
> > FILE:  elf-attrs.c line 232
> > FUNCTION: elf_new_obj_attr
> > /* Allocate/find an object attribute.  */
> > static obj_attribute *
> > elf_new_obj_attr (bfd *abfd, int vendor, unsigned int tag)
> > {
> >    obj_attribute *attr;
> >    obj_attribute_list *list;
> >    obj_attribute_list *p;
> >    obj_attribute_list **lastp;
> >
> >
> >    if (tag < NUM_KNOWN_OBJ_ATTRIBUTES)
> >      {
> >        /* Known tags are preallocated.  */
> >        attr = &elf_known_obj_attributes (abfd)[vendor][tag];
> >      }
> >    else
> >      {
> >        /* Create a new tag.  */
> >        list = (obj_attribute_list *)
> >       bfd_alloc (abfd, sizeof (obj_attribute_list));
> >        memset (list, 0, sizeof (obj_attribute_list));
> >        list->tag = tag;
> >        /* Keep the tag list in order.  */
> >        lastp = &elf_other_obj_attributes (abfd)[vendor];
> >        for (p = *lastp; p; p = p->next)
> >       {
> >         if (tag < p->tag)
> >           break;
> >         lastp = &p->next;
> >       }
> >        list->next = *lastp;
> >        *lastp = list;
> >        attr = &list->attr;
> >      }
> >
> >    return attr;
> > }
> >
> > bfd_alloc can return NULL. This is not checked
> > How to fix:
> > Add
> > if (list == NULL) return NULL;
>
> I've seen that you've sent a few mails similar to this one now.
> First of all, thanks for trying to improve the quality on binutils!
>
> If you want to help binutils even more, I would suggest that you create
> a proper patch and submit that instead of a mail with the above content
> as it would make it less work for a maintainer to actually commit your fix.
>
> Note, the above is just my personal opinion and might not be align with
> the preferences of the binutils maintainers.
>
> Kind regards,
> Torbjörn

I share the same opinion as Torbjorn's.

You might consider configuring  `git send-email` with a MTA (e.g.
msmtp), then you can send [PATCH] messages that a maintainer can apply
with ease:

    b4 am 'https://inbox.sourceware.org/binutils/DB9PR08MB647413BBA3340351A802035F81F1A@DB9PR08MB6474.eurprd08.prod.outlook.com/T/#t'
-o - | git am

(See https://josefbacik.github.io/kernel/2021/10/18/lei-and-b4.html)

If a patch is straightforward enough and a maintainer notices it, I am
sure this will make it faster for a fix to be applied.