[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20250603134738.GG386142@ziepe.ca>
Date: Tue, 3 Jun 2025 10:47:38 -0300
From: Jason Gunthorpe <jgg@...pe.ca>
To: Alistair Popple <apopple@...dia.com>
Cc: linux-mm@...ck.org, gerald.schaefer@...ux.ibm.com,
dan.j.williams@...el.com, willy@...radead.org, david@...hat.com,
linux-kernel@...r.kernel.org, nvdimm@...ts.linux.dev,
linux-fsdevel@...r.kernel.org, linux-ext4@...r.kernel.org,
linux-xfs@...r.kernel.org, jhubbard@...dia.com, hch@....de,
zhang.lyra@...il.com, debug@...osinc.com, bjorn@...nel.org,
balbirs@...dia.com, lorenzo.stoakes@...cle.com,
linux-arm-kernel@...ts.infradead.org, loongarch@...ts.linux.dev,
linuxppc-dev@...ts.ozlabs.org, linux-riscv@...ts.infradead.org,
linux-cxl@...r.kernel.org, dri-devel@...ts.freedesktop.org,
John@...ves.net
Subject: Re: [PATCH 06/12] mm/gup: Remove pXX_devmap usage from
get_user_pages()
On Thu, May 29, 2025 at 04:32:07PM +1000, Alistair Popple wrote:
> GUP uses pXX_devmap() calls to see if it needs to a get a reference on
> the associated pgmap data structure to ensure the pages won't go
> away. However it's a driver responsibility to ensure that if pages are
> mapped (ie. discoverable by GUP) that they are not offlined or removed
> from the memmap so there is no need to hold a reference on the pgmap
> data structure to ensure this.
Yes, the pgmap refcounting never made any sense here.
But I'm not sure this ever got fully fixed up?
To solve races with GUP fast we need a IPI/synchronize_rcu after all
VMAs are zapped and before the pgmap gets destroyed. Granted it is a
very small race in gup fast, it still should have this locking.
> Furthermore mappings with PFN_DEV are no longer created, hence this
> effectively dead code anyway so can be removed.
>
> Signed-off-by: Alistair Popple <apopple@...dia.com>
> ---
> include/linux/huge_mm.h | 3 +-
> mm/gup.c | 162 +----------------------------------------
> mm/huge_memory.c | 40 +----------
> 3 files changed, 5 insertions(+), 200 deletions(-)
Reviewed-by: Jason Gunthorpe <jgg@...dia.com>
Jason
Powered by blists - more mailing lists