Re: [PATCH v2] tools/mm: Fix slabinfo crash when MAX_SLABS is exceeded

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Tue, Oct 29, 2024 at 9:45 PM Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
>
> On Tue, 29 Oct 2024 13:13:41 -0300 Marc Dionne <marc.c.dionne@xxxxxxxxx> wrote:
>
> > From: Marc Dionne <marc.dionne@xxxxxxxxxxxx>
> >
> > The number of slabs can easily exceed the hard coded MAX_SLABS in the
> > slabinfo tool, causing it to overwrite memory and crash.
> >
> > Increase the value of MAX_SLABS, and check if that has been exceeded for
> > each new slab, instead of at the end when it's already too late.  Also
> > move the check for MAX_ALIASES into the loop body.
>
> Thanks.
>
> > --- a/tools/mm/slabinfo.c
> > +++ b/tools/mm/slabinfo.c
> > @@ -21,7 +21,7 @@
> >  #include <regex.h>
> >  #include <errno.h>
> >
> > -#define MAX_SLABS 500
> > +#define MAX_SLABS 1000
>
> That isn't a very large increase.

Fair point, given how quickly it has grown, maybe something like 4k
would give more headroom.

>
> >  #define MAX_ALIASES 500
> >  #define MAX_NODES 1024
> >
> > @@ -1240,6 +1240,8 @@ static void read_slab_dir(void)
> >                               p--;
> >                       alias->ref = strdup(p);
> >                       alias++;
> > +                     if (alias - aliasinfo > MAX_ALIASES)
> > +                             fatal("Too many aliases\n");
> >                       break;
> >                  case DT_DIR:
> >                       if (chdir(de->d_name))
> > @@ -1301,6 +1303,8 @@ static void read_slab_dir(void)
> >                       if (slab->name[0] == ':')
> >                               alias_targets++;
> >                       slab++;
> > +                     if (slab - slabinfo > MAX_SLABS)
> > +                             fatal("Too many slabs\n");
> >                       break;
>
> This could be improved - if the number of slabs is exactly equal to
> MAX_SLABS we'll unnecessarily report an error.  Wouldn't this
> alteration be better?

Actually the issue is rather that we'll accept MAX_SLABS + 1 when we
shouldn't, so yes the checks should be >=, or could also be just ==,
sorry.

>
> --- a/tools/mm/slabinfo.c~tools-mm-fix-slabinfo-crash-when-max_slabs-is-exceeded-fix
> +++ a/tools/mm/slabinfo.c
> @@ -1228,6 +1228,8 @@ static void read_slab_dir(void)
>                                 continue;
>                 switch (de->d_type) {
>                    case DT_LNK:
> +                       if (alias - aliasinfo >= MAX_ALIASES)
> +                               fatal("Too many aliases\n");
>                         alias->name = strdup(de->d_name);
>                         count = readlink(de->d_name, buffer, sizeof(buffer)-1);
>
> @@ -1240,10 +1242,10 @@ static void read_slab_dir(void)
>                                 p--;
>                         alias->ref = strdup(p);
>                         alias++;
> -                       if (alias - aliasinfo > MAX_ALIASES)
> -                               fatal("Too many aliases\n");
>                         break;
>                    case DT_DIR:
> +                       if (slab - slabinfo >= MAX_SLABS)
> +                               fatal("Too many slabs\n");
>                         if (chdir(de->d_name))
>                                 fatal("Unable to access slab %s\n", slab->name);
>                         slab->name = strdup(de->d_name);
> @@ -1305,8 +1307,6 @@ static void read_slab_dir(void)
>                         if (slab->name[0] == ':')
>                                 alias_targets++;
>                         slab++;
> -                       if (slab - slabinfo > MAX_SLABS)
> -                               fatal("Too many slabs\n");
>                         break;
>                    default :
>                         fatal("Unknown file type %lx\n", de->d_type);
> _
>





[Index of Archives]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Bugtraq]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux