linux-kernel - Re: [PATCH 5/5] kernel: tracepoints: add support for relative references

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 18 Aug 2017 09:36:31 +0100
From:   Ard Biesheuvel <ard.biesheuvel@...aro.org>
To:     Ingo Molnar <mingo@...nel.org>
Cc:     "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-arch@...r.kernel.org" <linux-arch@...r.kernel.org>,
        "H. Peter Anvin" <hpa@...or.com>, Arnd Bergmann <arnd@...db.de>,
        Heiko Carstens <heiko.carstens@...ibm.com>,
        Kees Cook <keescook@...omium.org>,
        Will Deacon <will.deacon@....com>,
        Michael Ellerman <mpe@...erman.id.au>,
        Thomas Garnier <thgarnie@...gle.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        "Serge E. Hallyn" <serge@...lyn.com>,
        Bjorn Helgaas <bhelgaas@...gle.com>,
        Benjamin Herrenschmidt <benh@...nel.crashing.org>,
        Paul Mackerras <paulus@...ba.org>,
        Catalin Marinas <catalin.marinas@....com>,
        Petr Mladek <pmladek@...e.com>, Ingo Molnar <mingo@...hat.com>,
        James Morris <james.l.morris@...cle.com>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Nicolas Pitre <nico@...aro.org>,
        Steven Rostedt <rostedt@...dmis.org>,
        Martin Schwidefsky <schwidefsky@...ibm.com>,
        Sergey Senozhatsky <sergey.senozhatsky@...il.com>,
        Jessica Yu <jeyu@...nel.org>
Subject: Re: [PATCH 5/5] kernel: tracepoints: add support for relative references

On 18 August 2017 at 09:26, Ingo Molnar <mingo@...nel.org> wrote:
>
> * Ard Biesheuvel <ard.biesheuvel@...aro.org> wrote:
>
>> -static void for_each_tracepoint_range(struct tracepoint * const *begin,
>> -             struct tracepoint * const *end,
>> +static void for_each_tracepoint_range(const void *begin, const void *end,
>>               void (*fct)(struct tracepoint *tp, void *priv),
>>               void *priv)
>>  {
>> +#ifdef CONFIG_HAVE_ARCH_PREL32_RELOCATIONS
>> +     const signed int *iter;
>> +
>> +     if (!begin)
>> +             return;
>> +     for (iter = begin; iter < (signed int *)end; iter++) {
>> +             fct((struct tracepoint *)((unsigned long)iter + *iter), priv);
>> +     }
>
> I think checkpatch is correct here to complain about the unnecessary curly braces
> here.
>

Fair enough. I will clean up to the extent feasible.

> Plus why the heavy use of 'signed int' here and elsewhere in the patches - why
> not 'int' ?
>

Yes, just 'int' should be sufficient. Force of habit, I suppose, given
that unqualified 'char' is ambiguous between architectures, but this
does not apply to 'int' of course.

> Plus #2, the heavy use of type casts looks pretty ugly to begin with - is there no
> way to turn this into more natural code?
>

Actually, Steven requested to keep the tracepoint section markers as
they are, and cast them to (int *) in the conditionally compiled
PREL32 case. So I guess it is a matter of taste, but because the types
are fundamentally incompatible when this code is in effect (and the
size of the pointed-to type differs on 64-bit architectures), there is
always some mangling required.

The initcall patch is probably the cleanest in this regard, but
involves typedefs, which are frowned upon as well.