From mboxrd@z Thu Jan  1 00:00:00 1970
From: Dave.Martin@arm.com (Dave Martin)
Date: Tue, 6 Jun 2017 14:58:18 +0100
Subject: [RFC PATCH v2 4/6] arm64: signal: Allocate extra sigcontext
 space as needed
In-Reply-To: <20170606113739.GF30160@e103592.cambridge.arm.com>
References: <1492016239-19511-1-git-send-email-Dave.Martin@arm.com>
 <1492016495-19689-1-git-send-email-Dave.Martin@arm.com>
 <1492016495-19689-4-git-send-email-Dave.Martin@arm.com>
 <20170512165724.GB8387@e104818-lin.cambridge.arm.com>
 <20170515132442.GB3559@e103592.cambridge.arm.com>
 <20170523113019.GB5948@e104818-lin.cambridge.arm.com>
 <20170526113729.GO3559@e103592.cambridge.arm.com>
 <20170605141744.GB9163@e104818-lin.cambridge.arm.com>
 <20170606113739.GF30160@e103592.cambridge.arm.com>
Message-ID: <20170606135815.GG30160@e103592.cambridge.arm.com>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

On Tue, Jun 06, 2017 at 12:37:53PM +0100, Dave Martin wrote:
> On Mon, Jun 05, 2017 at 03:17:44PM +0100, Catalin Marinas wrote:

[...]

[extra_context.size]

> > I'd rather have the time size_t or __u64 to avoid implicit padding.
> 
> Sure, it can be a u64.  I wanted to avoid the suggestion that the frame
> should be that large, but 32 bits already allows it to be crazy large
> anyway, so I don't think making it 32-bit helps.

Actually, there is still implicit padding even with u64, since the total
size is 16 bytes + sizeof(extra_context.size).

Since u64 is much bigger then we'd ever want, and to avoid introducing
new bugs, do you object to keeping size as u32 and adding explicit
padding instead?

Cheers
---Dave