From: Laurentiu Tudor <laurentiu.tudor@nxp.com>
To: Robin Murphy <robin.murphy@arm.com>,
"netdev@vger.kernel.org" <netdev@vger.kernel.org>,
Madalin-cristian Bucur <madalin.bucur@nxp.com>,
Roy Pledge <roy.pledge@nxp.com>,
Camelia Alexandra Groza <camelia.groza@nxp.com>,
Leo Li <leoyang.li@nxp.com>
Cc: "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
"iommu@lists.linux-foundation.org"
<iommu@lists.linux-foundation.org>,
"linuxppc-dev@lists.ozlabs.org" <linuxppc-dev@lists.ozlabs.org>,
"davem@davemloft.net" <davem@davemloft.net>,
"linux-arm-kernel@lists.infradead.org"
<linux-arm-kernel@lists.infradead.org>
Subject: RE: [PATCH 02/13] soc/fsl/bman: map FBPR area in the iommu
Date: Mon, 1 Apr 2019 11:04:52 +0000 [thread overview]
Message-ID: <DB7PR04MB51324E728B91663DECC6FD87EC550@DB7PR04MB5132.eurprd04.prod.outlook.com> (raw)
In-Reply-To: <b747c418-a495-56e0-c106-a8ec8e82ccb9@arm.com>
Hi Robin,
> -----Original Message-----
> From: Robin Murphy [mailto:robin.murphy@arm.com]
> Sent: Friday, March 29, 2019 4:51 PM
>
> On 29/03/2019 14:00, laurentiu.tudor@nxp.com wrote:
> > From: Laurentiu Tudor <laurentiu.tudor@nxp.com>
> >
> > Add a one-to-one iommu mapping for bman private data memory (FBPR).
> > This is required for BMAN to work without faults behind an iommu.
> >
> > Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
> > ---
> > drivers/soc/fsl/qbman/bman_ccsr.c | 11 +++++++++++
> > 1 file changed, 11 insertions(+)
> >
> > diff --git a/drivers/soc/fsl/qbman/bman_ccsr.c
> b/drivers/soc/fsl/qbman/bman_ccsr.c
> > index 7c3cc968053c..b209c79511bb 100644
> > --- a/drivers/soc/fsl/qbman/bman_ccsr.c
> > +++ b/drivers/soc/fsl/qbman/bman_ccsr.c
> > @@ -29,6 +29,7 @@
> > */
> >
> > #include "bman_priv.h"
> > +#include <linux/iommu.h>
> >
> > u16 bman_ip_rev;
> > EXPORT_SYMBOL(bman_ip_rev);
> > @@ -178,6 +179,7 @@ static int fsl_bman_probe(struct platform_device
> *pdev)
> > int ret, err_irq;
> > struct device *dev = &pdev->dev;
> > struct device_node *node = dev->of_node;
> > + struct iommu_domain *domain;
> > struct resource *res;
> > u16 id, bm_pool_cnt;
> > u8 major, minor;
> > @@ -225,6 +227,15 @@ static int fsl_bman_probe(struct platform_device
> *pdev)
> >
> > dev_dbg(dev, "Allocated FBPR 0x%llx 0x%zx\n", fbpr_a, fbpr_sz);
> >
> > + /* Create an 1-to-1 iommu mapping for FBPR area */
> > + domain = iommu_get_domain_for_dev(dev);
>
> If that's expected to be the default domain that you're grabbing, then
> this is *incredibly* fragile. There's nothing to stop the IOVA that you
> forcibly map from being automatically allocated later and causing some
> other DMA mapping to fail noisily and unexpectedly.
Agree here, we pretty much rely on luck with this implementation. As a side note, I've also experimented using dma_map_resource() instead of directly calling into iommu api and things worked fine, but see below ...
> Furthermore, have you tried this with "iommu.passthrough=1"?
Yes. The iommu_map() calls fail and the drivers issue warning messages, but apart from that I don't see any issues.
> That said, I really don't understand what's going on here anyway :/
>
> As far as I can tell from qbman_init_private_mem(), fbpr_a comes from
> dma_alloc_coherent() and thus would already be a mapped IOVA - isn't
> this the stuff that Roy converted to nicely use shared-dma-pool regions
> a while ago?
I must say that I'm also unclear on this. The thing is that I don't get to see a smmu mapping being created for the reserved memory as result of calling dma_alloc_coherent(). IIRC, at the time when I looked at this I concluded that the call to dma_alloc_coherent() simply returns the phys address of the device's reserved memory without creating a smmu mapping to back it up. Maybe my understanding was not correct or perhaps there's an issue with this shared-dma-pool mechanism where instead of creating a mapping in the smmu and return an IOVA it just returns the physical address of the reserved memory area.
---
Thanks & Best Regards, Laurentiu
>
> > + if (domain) {
> > + ret = iommu_map(domain, fbpr_a, fbpr_a, fbpr_sz,
> > + IOMMU_READ | IOMMU_WRITE | IOMMU_CACHE);
> > + if (ret)
> > + dev_warn(dev, "failed to iommu_map() %d\n", ret);
> > + }
> > +
> > bm_set_memory(fbpr_a, fbpr_sz);
> >
> > err_irq = platform_get_irq(pdev, 0);
> >
next prev parent reply other threads:[~2019-04-01 11:05 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-03-29 14:00 [PATCH 00/13] Prerequisites for NXP LS104xA SMMU enablement laurentiu.tudor
2019-03-29 14:00 ` [PATCH 01/13] soc/fsl/qman: fixup liodns only on ppc targets laurentiu.tudor
2019-03-29 21:50 ` Li Yang
2019-04-01 10:21 ` Laurentiu Tudor
2019-03-29 14:00 ` [PATCH 02/13] soc/fsl/bman: map FBPR area in the iommu laurentiu.tudor
2019-03-29 14:50 ` Robin Murphy
2019-04-01 11:04 ` Laurentiu Tudor [this message]
2019-04-19 13:41 ` Laurentiu Tudor
2019-03-29 21:16 ` Li Yang
2019-04-01 11:15 ` Laurentiu Tudor
2019-03-29 14:00 ` [PATCH 03/13] soc/fsl/qman: map FQD and PFDR areas " laurentiu.tudor
2019-03-29 14:00 ` [PATCH 04/13] soc/fsl/qman-portal: map CENA area " laurentiu.tudor
2019-03-29 14:00 ` [PATCH 05/13] soc/fsl/bqman: page align iommu mapping sizes laurentiu.tudor
2019-03-29 22:06 ` Li Yang
2019-04-01 10:18 ` Laurentiu Tudor
2019-03-29 14:00 ` [PATCH 06/13] soc/fsl/qbman_portals: add APIs to retrieve the probing status laurentiu.tudor
2019-04-01 22:12 ` Li Yang
2019-03-29 14:00 ` [PATCH 07/13] fsl/fman: backup and restore ICID registers laurentiu.tudor
2019-03-29 14:00 ` [PATCH 08/13] fsl/fman: add API to get the device behind a fman port laurentiu.tudor
2019-03-29 14:00 ` [PATCH 09/13] dpaa_eth: defer probing after qbman laurentiu.tudor
2019-03-29 14:00 ` [PATCH 10/13] dpaa_eth: base dma mappings on the fman rx port laurentiu.tudor
2019-03-29 14:00 ` [PATCH 11/13] dpaa_eth: fix iova handling for contiguous frames laurentiu.tudor
2019-03-29 14:00 ` [PATCH 12/13] dpaa_eth: fix iova handling for sg frames laurentiu.tudor
2019-03-29 14:00 ` [PATCH 13/13] dpaa_eth: fix SG frame cleanup laurentiu.tudor
2019-03-29 15:25 ` Joakim Tjernlund
2019-04-01 10:41 ` Laurentiu Tudor
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=DB7PR04MB51324E728B91663DECC6FD87EC550@DB7PR04MB5132.eurprd04.prod.outlook.com \
--to=laurentiu.tudor@nxp.com \
--cc=camelia.groza@nxp.com \
--cc=davem@davemloft.net \
--cc=iommu@lists.linux-foundation.org \
--cc=leoyang.li@nxp.com \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linuxppc-dev@lists.ozlabs.org \
--cc=madalin.bucur@nxp.com \
--cc=netdev@vger.kernel.org \
--cc=robin.murphy@arm.com \
--cc=roy.pledge@nxp.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®