Re: [PATCH] Revert arm64: drop ranges in definition of ARCH_FORCE_MAX_ORDER

From: Philip Li <philip.li@intel.com>
To: Catalin Marinas <catalin.marinas@arm.com>
Cc: Mike Rapoport <rppt@kernel.org>,
	Justin Forbes <jforbes@fedoraproject.org>,
	Will Deacon <will@kernel.org>,
	<linux-arm-kernel@lists.infradead.org>,
	<linux-kernel@vger.kernel.org>, <jmforbes@linuxtx.org>,
	Andrew Morton <akpm@linux-foundation.org>, <lkp@intel.com>
Subject: Re: [PATCH] Revert arm64: drop ranges in definition of ARCH_FORCE_MAX_ORDER
Date: Wed, 3 May 2023 20:08:51 +0800	[thread overview]
Message-ID: <ZFJO01C9jbZrmDMW@rli9-mobl> (raw)
In-Reply-To: <ZFI1eK5JmwIdL1eq@arm.com>

On Wed, May 03, 2023 at 11:20:40AM +0100, Catalin Marinas wrote:
> On Tue, May 02, 2023 at 06:40:14PM +0100, Catalin Marinas wrote:
> > On Tue, May 02, 2023 at 07:15:20PM +0300, Mike Rapoport wrote:
> > > On Tue, May 02, 2023 at 03:07:41PM +0100, Catalin Marinas wrote:
> > > > On Mon, May 01, 2023 at 04:24:38PM -0500, Justin Forbes wrote:
> > > > 
> > > > Regarding EXPERT, we could drop it and do like the other architectures
> > > > but we'll have randconfig occasionally hitting weird values that won't
> > > > build (like -1). Not sure EXPERT helps here.
> > > 
> > > AFAIU, randconfig does not randomize int values, it's probably random
> > > people that do ;-)
> > 
> > https://lore.kernel.org/r/202303232149.Chh6KhiI-lkp@intel.com
> > 
> > with the randconfig here:
> > 
> > https://download.01.org/0day-ci/archive/20230323/202303232149.Chh6KhiI-lkp@intel.com/config
> 
> You may be right, I can't get my randconfig to set ARCH_FORCE_MAX_ORDER
> to anything other than the default. Maybe the kernel test robot has its
> own config randomisation (cc'ing lkp@intel.com).

Really appologize here, around Mar 23 time period, there's a bug in kernel test
robot code which wrongly set the value of ARCH_FORCE_MAX_ORDER to -1. We fixed
it after noticing this, and replied to the false positive reports we sent out.
But we missed to reply this report to point out it was invalid report.

Sorry about this again, pls ignore this report.

> 
> If we don't care about about this randconfig, I'm fine do drop EXPERT
> from current mainline, together with the 4K/16K pages condition. The
> condition only made sense if we kept the ranges in since these were
> configurable (no range for 64K).
> 
> diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
> index b1201d25a8a4..1867aba83ba3 100644
> --- a/arch/arm64/Kconfig
> +++ b/arch/arm64/Kconfig
> @@ -1516,7 +1516,7 @@ config XEN
>  # 16K |       27          |      14      |       13        |         11         |
>  # 64K |       29          |      16      |       13        |         13         |
>  config ARCH_FORCE_MAX_ORDER
> -	int "Order of maximal physically contiguous allocations" if EXPERT && (ARM64_4K_PAGES || ARM64_16K_PAGES)
> +	int "Order of maximal physically contiguous allocations"
>  	default "13" if ARM64_64K_PAGES
>  	default "11" if ARM64_16K_PAGES
>  	default "10"
> 
> -- 
> Catalin

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel