From: "Jernej Škrabec" <jernej.skrabec@gmail.com>
To: joro@8bytes.org, will@kernel.org, Robin Murphy <robin.murphy@arm.com>
Cc: wens@csie.org, samuel@sholland.org, iommu@lists.linux.dev,
linux-arm-kernel@lists.infradead.org,
linux-sunxi@lists.linux.dev, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] iommu/sun50i: Allow page sizes between 4K and 1M
Date: Mon, 11 Sep 2023 21:29:18 +0200 [thread overview]
Message-ID: <5707055.DvuYhMxLoT@archlinux> (raw)
In-Reply-To: <40649946-739b-fec6-d743-f7828613fec5@arm.com>
On Monday, September 11, 2023 9:06:12 PM CEST Robin Murphy wrote:
> On 2023-09-11 20:02, Jernej Skrabec wrote:
> > While peripheral supports only 4K page sizes, we can easily emulate
> > support for bigger page sizes, up to 1M. This is done by making multiple
> > entries in map function or clearing multiple entries in unmap.
>
> No, it's done by converting over to the {map,unmap}_pages ops. There's
> no need to resurrect this horrible bodge.
Missed those, thanks! In any case, logic to implement those isn't that
different than here.
Best regards,
Jernej
>
> Thanks,
> Robin.
>
> > Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
> > ---
> >
> > drivers/iommu/sun50i-iommu.c | 49 +++++++++++++++++++++++-------------
> > 1 file changed, 31 insertions(+), 18 deletions(-)
> >
> > diff --git a/drivers/iommu/sun50i-iommu.c b/drivers/iommu/sun50i-iommu.c
> > index 74c5cb93e900..93077575d60f 100644
> > --- a/drivers/iommu/sun50i-iommu.c
> > +++ b/drivers/iommu/sun50i-iommu.c
> > @@ -95,6 +95,10 @@
> >
> > #define SPAGE_SIZE 4096
> >
> > +#define SUN50I_IOMMU_PGSIZES (SZ_4K | SZ_8K | SZ_16K | SZ_32K |
\
> > + SZ_64K | SZ_128K |
SZ_256K | \
> > + SZ_512K | SZ_1M)
> > +
> >
> > struct sun50i_iommu {
> >
> > struct iommu_device iommu;
> >
> > @@ -593,10 +597,12 @@ static int sun50i_iommu_map(struct iommu_domain
> > *domain, unsigned long iova,>
> > {
> >
> > struct sun50i_iommu_domain *sun50i_domain =
to_sun50i_domain(domain);
> > struct sun50i_iommu *iommu = sun50i_domain->iommu;
> >
> > - u32 pte_index;
> > + u32 pte_index, pages, i;
> >
> > u32 *page_table, *pte_addr;
> > int ret = 0;
> >
> > + pages = size / SPAGE_SIZE;
> > +
> >
> > page_table = sun50i_dte_get_page_table(sun50i_domain, iova, gfp);
> > if (IS_ERR(page_table)) {
> >
> > ret = PTR_ERR(page_table);
> >
> > @@ -604,18 +610,22 @@ static int sun50i_iommu_map(struct iommu_domain
> > *domain, unsigned long iova,>
> > }
> >
> > pte_index = sun50i_iova_get_pte_index(iova);
> >
> > - pte_addr = &page_table[pte_index];
> > - if (unlikely(sun50i_pte_is_page_valid(*pte_addr))) {
> > - phys_addr_t page_phys =
sun50i_pte_get_page_address(*pte_addr);
> > - dev_err(iommu->dev,
> > - "iova %pad already mapped to %pa cannot
remap to %pa prot: %#x\n",
> > - &iova, &page_phys, &paddr, prot);
> > - ret = -EBUSY;
> > - goto out;
> > + for (i = 0; i < pages; i++) {
> > + pte_addr = &page_table[pte_index + i];
> > + if (unlikely(sun50i_pte_is_page_valid(*pte_addr))) {
> > + phys_addr_t page_phys =
sun50i_pte_get_page_address(*pte_addr);
> > +
> > + dev_err(iommu->dev,
> > + "iova %pad already mapped to %pa
cannot remap to %pa prot: %#x\n",
> > + &iova, &page_phys, &paddr, prot);
> > + ret = -EBUSY;
> > + goto out;
> > + }
> > + *pte_addr = sun50i_mk_pte(paddr, prot);
> > + paddr += SPAGE_SIZE;
> >
> > }
> >
> > - *pte_addr = sun50i_mk_pte(paddr, prot);
> > - sun50i_table_flush(sun50i_domain, pte_addr, 1);
> > + sun50i_table_flush(sun50i_domain, &page_table[pte_index], pages);
> >
> > out:
> > return ret;
> >
> > @@ -626,8 +636,10 @@ static size_t sun50i_iommu_unmap(struct iommu_domain
> > *domain, unsigned long iova>
> > {
> >
> > struct sun50i_iommu_domain *sun50i_domain =
to_sun50i_domain(domain);
> > phys_addr_t pt_phys;
> >
> > + u32 dte, pages, i;
> >
> > u32 *pte_addr;
> >
> > - u32 dte;
> > +
> > + pages = size / SPAGE_SIZE;
> >
> > dte = sun50i_domain->dt[sun50i_iova_get_dte_index(iova)];
> > if (!sun50i_dte_is_pt_valid(dte))
> >
> > @@ -636,13 +648,14 @@ static size_t sun50i_iommu_unmap(struct iommu_domain
> > *domain, unsigned long iova>
> > pt_phys = sun50i_dte_get_pt_address(dte);
> > pte_addr = (u32 *)phys_to_virt(pt_phys) +
> > sun50i_iova_get_pte_index(iova);
> >
> > - if (!sun50i_pte_is_page_valid(*pte_addr))
> > - return 0;
> > + for (i = 0; i < pages; i++)
> > + if (!sun50i_pte_is_page_valid(pte_addr[i]))
> > + return 0;
> >
> > - memset(pte_addr, 0, sizeof(*pte_addr));
> > - sun50i_table_flush(sun50i_domain, pte_addr, 1);
> > + memset(pte_addr, 0, sizeof(*pte_addr) * pages);
> > + sun50i_table_flush(sun50i_domain, pte_addr, pages);
> >
> > - return SZ_4K;
> > + return size;
> >
> > }
> >
> > static phys_addr_t sun50i_iommu_iova_to_phys(struct iommu_domain
> > *domain,
> >
> > @@ -827,7 +840,7 @@ static int sun50i_iommu_of_xlate(struct device *dev,
> >
> > }
> >
> > static const struct iommu_ops sun50i_iommu_ops = {
> >
> > - .pgsize_bitmap = SZ_4K,
> > + .pgsize_bitmap = SUN50I_IOMMU_PGSIZES,
> >
> > .device_group = sun50i_iommu_device_group,
> > .domain_alloc = sun50i_iommu_domain_alloc,
> > .of_xlate = sun50i_iommu_of_xlate,
WARNING: multiple messages have this Message-ID (diff)
From: "Jernej Škrabec" <jernej.skrabec@gmail.com>
To: joro@8bytes.org, will@kernel.org, Robin Murphy <robin.murphy@arm.com>
Cc: wens@csie.org, samuel@sholland.org, iommu@lists.linux.dev,
linux-arm-kernel@lists.infradead.org,
linux-sunxi@lists.linux.dev, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] iommu/sun50i: Allow page sizes between 4K and 1M
Date: Mon, 11 Sep 2023 21:29:18 +0200 [thread overview]
Message-ID: <5707055.DvuYhMxLoT@archlinux> (raw)
In-Reply-To: <40649946-739b-fec6-d743-f7828613fec5@arm.com>
On Monday, September 11, 2023 9:06:12 PM CEST Robin Murphy wrote:
> On 2023-09-11 20:02, Jernej Skrabec wrote:
> > While peripheral supports only 4K page sizes, we can easily emulate
> > support for bigger page sizes, up to 1M. This is done by making multiple
> > entries in map function or clearing multiple entries in unmap.
>
> No, it's done by converting over to the {map,unmap}_pages ops. There's
> no need to resurrect this horrible bodge.
Missed those, thanks! In any case, logic to implement those isn't that
different than here.
Best regards,
Jernej
>
> Thanks,
> Robin.
>
> > Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
> > ---
> >
> > drivers/iommu/sun50i-iommu.c | 49 +++++++++++++++++++++++-------------
> > 1 file changed, 31 insertions(+), 18 deletions(-)
> >
> > diff --git a/drivers/iommu/sun50i-iommu.c b/drivers/iommu/sun50i-iommu.c
> > index 74c5cb93e900..93077575d60f 100644
> > --- a/drivers/iommu/sun50i-iommu.c
> > +++ b/drivers/iommu/sun50i-iommu.c
> > @@ -95,6 +95,10 @@
> >
> > #define SPAGE_SIZE 4096
> >
> > +#define SUN50I_IOMMU_PGSIZES (SZ_4K | SZ_8K | SZ_16K | SZ_32K |
\
> > + SZ_64K | SZ_128K |
SZ_256K | \
> > + SZ_512K | SZ_1M)
> > +
> >
> > struct sun50i_iommu {
> >
> > struct iommu_device iommu;
> >
> > @@ -593,10 +597,12 @@ static int sun50i_iommu_map(struct iommu_domain
> > *domain, unsigned long iova,>
> > {
> >
> > struct sun50i_iommu_domain *sun50i_domain =
to_sun50i_domain(domain);
> > struct sun50i_iommu *iommu = sun50i_domain->iommu;
> >
> > - u32 pte_index;
> > + u32 pte_index, pages, i;
> >
> > u32 *page_table, *pte_addr;
> > int ret = 0;
> >
> > + pages = size / SPAGE_SIZE;
> > +
> >
> > page_table = sun50i_dte_get_page_table(sun50i_domain, iova, gfp);
> > if (IS_ERR(page_table)) {
> >
> > ret = PTR_ERR(page_table);
> >
> > @@ -604,18 +610,22 @@ static int sun50i_iommu_map(struct iommu_domain
> > *domain, unsigned long iova,>
> > }
> >
> > pte_index = sun50i_iova_get_pte_index(iova);
> >
> > - pte_addr = &page_table[pte_index];
> > - if (unlikely(sun50i_pte_is_page_valid(*pte_addr))) {
> > - phys_addr_t page_phys =
sun50i_pte_get_page_address(*pte_addr);
> > - dev_err(iommu->dev,
> > - "iova %pad already mapped to %pa cannot
remap to %pa prot: %#x\n",
> > - &iova, &page_phys, &paddr, prot);
> > - ret = -EBUSY;
> > - goto out;
> > + for (i = 0; i < pages; i++) {
> > + pte_addr = &page_table[pte_index + i];
> > + if (unlikely(sun50i_pte_is_page_valid(*pte_addr))) {
> > + phys_addr_t page_phys =
sun50i_pte_get_page_address(*pte_addr);
> > +
> > + dev_err(iommu->dev,
> > + "iova %pad already mapped to %pa
cannot remap to %pa prot: %#x\n",
> > + &iova, &page_phys, &paddr, prot);
> > + ret = -EBUSY;
> > + goto out;
> > + }
> > + *pte_addr = sun50i_mk_pte(paddr, prot);
> > + paddr += SPAGE_SIZE;
> >
> > }
> >
> > - *pte_addr = sun50i_mk_pte(paddr, prot);
> > - sun50i_table_flush(sun50i_domain, pte_addr, 1);
> > + sun50i_table_flush(sun50i_domain, &page_table[pte_index], pages);
> >
> > out:
> > return ret;
> >
> > @@ -626,8 +636,10 @@ static size_t sun50i_iommu_unmap(struct iommu_domain
> > *domain, unsigned long iova>
> > {
> >
> > struct sun50i_iommu_domain *sun50i_domain =
to_sun50i_domain(domain);
> > phys_addr_t pt_phys;
> >
> > + u32 dte, pages, i;
> >
> > u32 *pte_addr;
> >
> > - u32 dte;
> > +
> > + pages = size / SPAGE_SIZE;
> >
> > dte = sun50i_domain->dt[sun50i_iova_get_dte_index(iova)];
> > if (!sun50i_dte_is_pt_valid(dte))
> >
> > @@ -636,13 +648,14 @@ static size_t sun50i_iommu_unmap(struct iommu_domain
> > *domain, unsigned long iova>
> > pt_phys = sun50i_dte_get_pt_address(dte);
> > pte_addr = (u32 *)phys_to_virt(pt_phys) +
> > sun50i_iova_get_pte_index(iova);
> >
> > - if (!sun50i_pte_is_page_valid(*pte_addr))
> > - return 0;
> > + for (i = 0; i < pages; i++)
> > + if (!sun50i_pte_is_page_valid(pte_addr[i]))
> > + return 0;
> >
> > - memset(pte_addr, 0, sizeof(*pte_addr));
> > - sun50i_table_flush(sun50i_domain, pte_addr, 1);
> > + memset(pte_addr, 0, sizeof(*pte_addr) * pages);
> > + sun50i_table_flush(sun50i_domain, pte_addr, pages);
> >
> > - return SZ_4K;
> > + return size;
> >
> > }
> >
> > static phys_addr_t sun50i_iommu_iova_to_phys(struct iommu_domain
> > *domain,
> >
> > @@ -827,7 +840,7 @@ static int sun50i_iommu_of_xlate(struct device *dev,
> >
> > }
> >
> > static const struct iommu_ops sun50i_iommu_ops = {
> >
> > - .pgsize_bitmap = SZ_4K,
> > + .pgsize_bitmap = SUN50I_IOMMU_PGSIZES,
> >
> > .device_group = sun50i_iommu_device_group,
> > .domain_alloc = sun50i_iommu_domain_alloc,
> > .of_xlate = sun50i_iommu_of_xlate,
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
next prev parent reply other threads:[~2023-09-11 19:29 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-09-11 19:02 [PATCH] iommu/sun50i: Allow page sizes between 4K and 1M Jernej Skrabec
2023-09-11 19:02 ` Jernej Skrabec
2023-09-11 19:06 ` Robin Murphy
2023-09-11 19:06 ` Robin Murphy
2023-09-11 19:29 ` Jernej Škrabec [this message]
2023-09-11 19:29 ` Jernej Škrabec
2023-09-11 22:00 ` Robin Murphy
2023-09-11 22:00 ` Robin Murphy
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=5707055.DvuYhMxLoT@archlinux \
--to=jernej.skrabec@gmail.com \
--cc=iommu@lists.linux.dev \
--cc=joro@8bytes.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-sunxi@lists.linux.dev \
--cc=robin.murphy@arm.com \
--cc=samuel@sholland.org \
--cc=wens@csie.org \
--cc=will@kernel.org \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.