linux-kernel - Re: [PATCH 8/9] RISC-V: User-facing API

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]

Message-ID: <alpine.DEB.2.20.1706282336270.1890@nanos>
Date:   Wed, 28 Jun 2017 23:49:44 +0200 (CEST)
From:   Thomas Gleixner <tglx@...utronix.de>
To:     Palmer Dabbelt <palmer@...belt.com>
cc:     peterz@...radead.org, mingo@...hat.com, mcgrof@...nel.org,
        viro@...iv.linux.org.uk, sfr@...b.auug.org.au,
        nicolas.dichtel@...nd.com, rmk+kernel@...linux.org.uk,
        msalter@...hat.com, tklauser@...tanz.ch, will.deacon@....com,
        james.hogan@...tec.com, paul.gortmaker@...driver.com,
        linux@...ck-us.net, linux-kernel@...r.kernel.org,
        linux-arch@...r.kernel.org, albert@...ive.com
Subject: Re: [PATCH 8/9] RISC-V: User-facing API

On Wed, 28 Jun 2017, Palmer Dabbelt wrote:
> +
> +SYSCALL_DEFINE3(sysriscv_cmpxchg32, unsigned long, arg1, unsigned long, arg2,
> +		unsigned long, arg3)
> +{
> +	unsigned long flags;
> +	unsigned long prev;
> +	unsigned int *ptr;
> +	unsigned int err;
> +
> +	ptr = (unsigned int *)arg1;

Errm. Why isn't arg1 a proper pointer type and the arguments arg2/3 u32?

And please give the arguments a proper name, so it's obvious what is what.

SYSCALL_DEFINE3(sysriscv_cmpxchg32, u32 __user *, ptr, u32 new, u32 old)

Hmm?

> +	if (!access_ok(VERIFY_WRITE, ptr, sizeof(unsigned int)))
> +		return -EFAULT;
> +
> +	preempt_disable();
> +	raw_local_irq_save(flags);

Why do you want to disable interrupts here? This is thread context and
accessing user space memory, so the only protection this needs is against
preemption.

> +	err = __get_user(prev, ptr);
> +	if (likely(!err && prev == arg2))
> +		err = __put_user(arg3, ptr);
> +	raw_local_irq_restore(flags);
> +	preempt_enable();
> +
> +	return unlikely(err) ? err : prev;
> +}
> +
> +SYSCALL_DEFINE3(sysriscv_cmpxchg64, unsigned long, arg1, unsigned long, arg2,
> +		unsigned long, arg3)

This one is even worse. How does this implement cmpxchg64 on a 32bit machine?

Answer: Not at all, because arg2 and 3 are 32bit ....

> +{
> +	unsigned long flags;
> +	unsigned long prev;
> +	unsigned int *ptr;
> +	unsigned int err;
> +
> +	ptr = (unsigned int *)arg1;

Type casting to random pointer types makes the code more obvious
and safe, right? What the heck has a int pointer to do with u64?

> +	if (!access_ok(VERIFY_WRITE, ptr, sizeof(unsigned long)))
> +		return -EFAULT;
> +
> +	preempt_disable();
> +	raw_local_irq_save(flags);

Same as above.

> +	err = __get_user(prev, ptr);

Sigh. Type safety is overrated, right?

Thanks,

	tglx