linux-kernel - Re: [PATCH 16/17] riscv: clear the instruction cache and all registers when booting

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] [day] [month] [year] [list]

Date:   Tue, 13 Aug 2019 17:40:08 +0200
From:   Christoph Hellwig <hch@....de>
To:     Palmer Dabbelt <palmer@...ive.com>
Cc:     Atish Patra <Atish.Patra@....com>, Christoph Hellwig <hch@....de>,
        Paul Walmsley <paul.walmsley@...ive.com>, linux-mm@...ck.org,
        Damien Le Moal <Damien.LeMoal@....com>,
        linux-riscv@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 16/17] riscv: clear the instruction cache and all
 registers when booting

On Mon, Jul 08, 2019 at 01:26:33AM -0700, Palmer Dabbelt wrote:
>>> +	csrs	sstatus, t1
>
> You need to check that the write stuck and branch around the FP instructions.
> Specifically, CONFIG_FPU means there may be an FPU, not there's definately an
> FPU.  You should also turn the FPU back off after zeroing the state.

Well, that is why we check the hwcaps from misa just above and skip
this fp reg clearing if it doesn't contain the 'F' or 'D' extension.

The caller disables the FPU a few instructions later:

	/*
         * Disable FPU to detect illegal usage of
	 * floating point in kernel space
	 */
	li t0, SR_FS
	csrc CSR_XSTATUS, t0