Received-SPF: pass (google.com: domain of linux-kernel+bounces-170150-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
Date: Mon, 6 May 2024 09:18:28 -0700
From: Christoph Hellwig <hch@infradead.org>
To: Rick Edgecombe <rick.p.edgecombe@intel.com>
Cc: akpm@linux-foundation.org, Liam.Howlett@oracle.com, bp@alien8.de,
	bpf@vger.kernel.org, broonie@kernel.org,
	christophe.leroy@csgroup.eu, dan.j.williams@intel.com,
	dave.hansen@linux.intel.com, debug@rivosinc.com, hpa@zytor.com,
	io-uring@vger.kernel.org, keescook@chromium.org,
	kirill.shutemov@linux.intel.com, linux-cxl@vger.kernel.org,
	linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-mm@kvack.org, linux-s390@vger.kernel.org,
	linux-sgx@vger.kernel.org, luto@kernel.org, mingo@redhat.com,
	nvdimm@lists.linux.dev, peterz@infradead.org,
	sparclinux@vger.kernel.org, tglx@linutronix.de, x86@kernel.org
Subject: Re: [PATCH] mm: Remove mm argument from mm_get_unmapped_area()
Message-ID: <ZjkC1KgTTbR4qzhC@infradead.org>
References: <20240506160747.1321726-1-rick.p.edgecombe@intel.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20240506160747.1321726-1-rick.p.edgecombe@intel.com>

On Mon, May 06, 2024 at 09:07:47AM -0700, Rick Edgecombe wrote:
>  	if (flags & MAP_FIXED) {
>  		/* Ok, don't mess with it. */
> -		return mm_get_unmapped_area(current->mm, NULL, orig_addr, len, pgoff, flags);
> +		return current_get_unmapped_area(NULL, orig_addr, len, pgoff, flags);

The old name seems preferable because it's not as crazy long.  In fact
just get_unmapped_area would be even better, but that's already taken
by something else.

Can we maybe take a step back and sort out the mess of the various
_get_unmapped_area helpers?

e.g. mm_get_unmapped_area_vmflags just wraps
arch_get_unmapped_area_topdown_vmflags and
arch_get_unmapped_area_vmflags, and we might as well merge all three
by moving the MMF_TOPDOWN into two actual implementations?

And then just update all the implementations to always pass the
vm_flags instead of having separate implementations with our without
the flags.

And then make __get_unmapped_area static in mmap.c nad move the
get_unmapped_area wrappers there.  And eventually write some
documentation for the functions based on the learnings who actually
uses what..