Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Fri, 5 Nov 2021 09:56:16 -0700
From:   Ira Weiny <ira.weiny@intel.com>
To:     Matthew Wilcox <willy@infradead.org>
Cc:     Andrew Morton <akpm@linux-foundation.org>,
        Prathu Baronia <prathubaronia2011@gmail.com>,
        linux-kernel@vger.kernel.org, chintan.pandya@oneplus.com,
        Prathu Baronia <prathu.baronia@oneplus.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        "Peter Zijlstra (Intel)" <peterz@infradead.org>,
        Randy Dunlap <rdunlap@infradead.org>
Subject: Re: [PATCH v4 1/1] mm/highmem: Remove deprecated kmap_atomic
Message-ID: <20211105165616.GZ3538886@iweiny-DESK2.sc.intel.com>
References: <20210204073255.20769-1-prathu.baronia@oneplus.com>
 <20210204073255.20769-2-prathu.baronia@oneplus.com>
 <20210211003307.GA3158182@iweiny-DESK2.sc.intel.com>
 <20210211155625.7f8dd5d53e19f88530ddf248@linux-foundation.org>
 <YYU2YSOGHJFIrzmm@casper.infradead.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <YYU2YSOGHJFIrzmm@casper.infradead.org>
User-Agent: Mutt/1.11.1 (2018-12-01)
Precedence: bulk

On Fri, Nov 05, 2021 at 01:49:21PM +0000, Matthew Wilcox wrote:
> On Thu, Feb 11, 2021 at 03:56:25PM -0800, Andrew Morton wrote:
> > On Wed, 10 Feb 2021 16:33:07 -0800 Ira Weiny <ira.weiny@intel.com> wrote:
> > 
> > > > Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> > > 
> > > This already has my signed off by so I'm not going to 'review'.  With Prathu's
> > > testing information I hope this can land.
> > > 
> > > Andrew did you see this patch?
> > 
> > I did now ;)
> > 
> > Tossed onto the post-rc1 pile, thanks,
> 
> This patch seems to have slipped through the gaps for a couple of cycles
> now?  I found a missed spot in it for CONFIG_HIGHMEM:
> 
> diff --git a/mm/highmem.c b/mm/highmem.c
> index 471d9779a7f4..82d8c5ab6e8d 100644
> --- a/mm/highmem.c
> +++ b/mm/highmem.c
> @@ -382,7 +382,7 @@ void zero_user_segments(struct page *page, unsigned start1, unsigned end1,
>  			unsigned this_end = min_t(unsigned, end1, PAGE_SIZE);
>  
>  			if (end1 > start1) {
> -				kaddr = kmap_atomic(page + i);
> +				kaddr = kmap_local(page + i);

kmap_local_page()

>  				memset(kaddr + start1, 0, this_end - start1);
>  			}
>  			end1 -= this_end;
> @@ -397,7 +397,7 @@ void zero_user_segments(struct page *page, unsigned start1, unsigned end1,
>  
>  			if (end2 > start2) {
>  				if (!kaddr)
> -					kaddr = kmap_atomic(page + i);
> +					kaddr = kmap_local(page + i);

kmap_local_page()

I'm ok with this.  I'm not sure if Prathu needed this or not.

Also I wonder if memset_page could be used?  It would end up mapping the page
2x sometimes.

As an aside I think flush_dcache_page() needs to be in memset_page() for
completeness but I'm a bit afraid of adding it with the current controversy...
:-/

Anyway, why hasn't Brathu's patch[1] landed yet?

I just checked 5.15 and it is not there?  I don't even see it in the 5.16
merge?  Perhaps it is coming in this merge window?  Andrew?

Ira

[1] https://lore.kernel.org/lkml/20210204073255.20769-2-prathu.baronia@oneplus.com/

>  				memset(kaddr + start2, 0, this_end - start2);
>  			}
>  			end2 -= this_end;
> @@ -405,7 +405,7 @@ void zero_user_segments(struct page *page, unsigned start1, unsigned end1,
>  		}
>  
>  		if (kaddr) {
> -			kunmap_atomic(kaddr);
> +			kunmap_local(kaddr);
>  			flush_dcache_page(page + i);
>  		}