From mboxrd@z Thu Jan  1 00:00:00 1970
From: yangyingliang@huawei.com (Yang Yingliang)
Date: Wed, 8 Jun 2016 17:18:26 +0800
Subject: [RFC PATCH 0/4] arm64: vdso: check whether the params is invalid
 address
In-Reply-To: <20160606171248.GL669@arm.com>
References: <1464750362-14188-1-git-send-email-yangyingliang@huawei.com>
 <20160606171248.GL669@arm.com>
Message-ID: <5757E2E2.3010609@huawei.com>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org



On 2016/6/7 1:12, Will Deacon wrote:
> On Wed, Jun 01, 2016 at 11:05:58AM +0800, Yang Yingliang wrote:
>> The params of gettimeofday(), clock_gettime() and clock_getres()
>> is not checked. This will cause segment faults when the address is
>> invalid.
>> E.g. gettimeofday(-1, -1);
>>       clock_gettime(0, -1);
>>       clock_getres(0, -1);
>> This patchset add a macro and use it to check the validation of these
>> pointer params.
>
> Is this really necessary? It doesn't look to me like other architectures
> bother with this, and you're just adding code to the fastpath to handle
> cases that shouldn't happen to start with.
I agree.

I find clock_gettime() don't check if tp is NULL. I think it need a
checking like clock_getres().

What's your suggestion ?

Thanks,
Yang
>
> Also, I think I remember seeing SEGVs from libc wrappers for other syscalls
> in the past, so this just isn't compelling to me.
>
> Will
>
> .
>