Re: KVM guest sometimes failed to boot because of kernel stack overflow if KPTI is enabled on a hisilicon ARM64 platform.

From: Will Deacon <will.deacon@arm.com>
To: Wei Xu <xuwei5@hisilicon.com>
Cc: James Morse <james.morse@arm.com>,
	catalin.marinas@arm.com, suzuki.poulose@arm.com,
	dave.martin@arm.com, mark.rutland@arm.com, marc.zyngier@arm.com,
	linux-arm-kernel@lists.infradead.org,
	linux-kernel@vger.kernel.org, Linuxarm <linuxarm@huawei.com>,
	Hanjun Guo <guohanjun@huawei.com>,
	xiexiuqi@huawei.com, huangdaode <huangdaode@hisilicon.com>,
	"Chenxin (Charles)" <charles.chenxin@huawei.com>,
	"Xiongfanggou (James)" <james.xiong@huawei.com>,
	"Liguozhu (Kenneth)" <liguozhu@hisilicon.com>,
	Zhangyi ac <zhangyi.ac@huawei.com>,
	jonathan.cameron@huawei.com,
	Shameerali Kolothum Thodi  <shameerali.kolothum.thodi@huawei.com>,
	John Garry <john.garry@huawei.com>,
	Salil Mehta <salil.mehta@huawei.com>,
	Shiju Jose <shiju.jose@huawei.com>,
	"Zhuangyuzeng (Yisen)" <yisen.zhuang@huawei.com>,
	"Wangzhou (B)" <wangzhou1@hisilicon.com>,
	"kongxinwei (A)" <kong.kongxinwei@hisilicon.com>,
	"Liyuan (Larry, Turing Solution)" <Larry.T@huawei.com>,
	libeijian@hisilicon.com, zhangbin011@hisilicon.com
Subject: Re: KVM guest sometimes failed to boot because of kernel stack overflow if KPTI is enabled on a hisilicon ARM64 platform.
Date: Fri, 22 Jun 2018 14:31:34 +0100	[thread overview]
Message-ID: <20180622133133.GA1802@arm.com> (raw)
In-Reply-To: <5B2CF723.7010600@hisilicon.com>

Hi again, Wei,

On Fri, Jun 22, 2018 at 09:18:27PM +0800, Wei Xu wrote:
> On 2018/6/22 19:16, Will Deacon wrote:
> >On Fri, Jun 22, 2018 at 06:45:15PM +0800, Wei Xu wrote:
> >>On 2018/6/22 17:23, Will Deacon wrote:
> >>>Perhaps just writing back the table entries is enough to cause the issue,
> >>>although I really can't understand why that would be the case. Can you try
> >>>the diff below (without my previous change), please?
> >>Thanks!
> >>But it does not resolve the issue(only apply this patch based on 4.17.0).
> >Thanks, that's a useful data point. It means that it still crashes even if
> >we write back the same table entries, so it's the fact that we're writing
> >them at all which causes the problem, not the value that we write.
> >
> >Whilst looking at the code, we noticed a missing DMB. On the off-chance
> >that it helps, can you try this instead please?
> Thanks!
> Only apply below patch based on 4.17.0, we still got the crash.

Oh well, it was worth a shot (and that's still a fix worth having). Please
can you provide the complete disassembly for kpti_install_ng_mappings()
(I'm referring to the C function in cpufeature.c) along with a corresponding
crash log so that we can correlate the instruction stream with the crash?

Thanks,

Will