[Top][All Lists]

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

Re: [PATCH v5 05/22] target/arm: Suppress tag check for sp+offset

From: Peter Maydell
Subject: Re: [PATCH v5 05/22] target/arm: Suppress tag check for sp+offset
Date: Tue, 3 Dec 2019 14:07:44 +0000

On Fri, 11 Oct 2019 at 14:49, Richard Henderson
<address@hidden> wrote:
> R0078 specifies that base register, or base register plus immediate
> offset, is unchecked when the base register is SP.

It looks like rule-numbers didn't make it into the final Arm ARM,
so I guess the reference here would just be to section D6.8.1 ?

Also, this phrasing is slightly ambiguous about whether the
"when base is SP" condition applies to both "base register"
and "base register + immediate", or just to the last of the two;
the correct reading is the latter of these (and the D6.8.1
Arm ARM text is in error; trust the pseudocode here).

We could perhaps say something like:

D6.8.1 specifies that accesses are tag-unchecked for loads and
stores (including exclusives, compare-and-swap, etc) whose addresses are:
 * base-register only, where the base register is SP
 * base-register plus immediate, where the base register is SP
   (not including reg+imm with writeback addressing forms)
and also that literal (pc-relative) loads are tag-unchecked.

> Signed-off-by: Richard Henderson <address@hidden>
> ---
> v2: Include writeback addresses as checked.

The load-literal case is implicitly tag-unchecked because
the address calculation doesn't go via clean_data_tbi(), right?

Reviewed-by: Peter Maydell <address@hidden>

-- PMM

reply via email to

[Prev in Thread] Current Thread [Next in Thread]