[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ZFPuB8+Kb0lvI/yx@nvidia.com>
Date: Thu, 4 May 2023 14:40:23 -0300
From: Jason Gunthorpe <jgg@...dia.com>
To: Brett Creeley <brett.creeley@....com>
Cc: kvm@...r.kernel.org, netdev@...r.kernel.org, alex.williamson@...hat.com,
yishaih@...dia.com, shameerali.kolothum.thodi@...wei.com,
kevin.tian@...el.com, shannon.nelson@....com, drivers@...sando.io
Subject: Re: [PATCH v9 vfio 4/7] vfio/pds: Add VFIO live migration support
On Fri, Apr 21, 2023 at 06:06:39PM -0700, Brett Creeley wrote:
> +static struct pds_vfio_lm_file *
> +pds_vfio_get_lm_file(const struct file_operations *fops, int flags, u64 size)
> +{
I see this function is called with a hardwired 64k size on restore -
so this means save is also limited to 64k?
This is really overcomplicated if that is the state size you are
working with - the mlx5 code this was copied from is dealing with much
larger values.
Just kvalloc() your 64k and be done with it. You need a bit of fussing
to DMA map the vmap, but it is much simpler than all of this
stuff. See fpga_mgr_buf_load()
Since the kvalloc is linear the read/write is just memcpy.
Jason
Powered by blists - more mailing lists