diff mbox series

[net-next,v5,05/14] net: fman: Map the base address once

Message ID 20220902215737.981341-6-sean.anderson@seco.com (mailing list archive)
State New, archived
Headers show
Series net: dpaa: Cleanups in preparation for phylink conversion (part 2) | expand

Commit Message

Sean Anderson Sept. 2, 2022, 9:57 p.m. UTC
We don't need to remap the base address from the resource twice (once in
mac_probe() and again in set_fman_mac_params()). We still need the
resource to get the end address, but we can use a single function call
to get both at once.

While we're at it, use platform_get_mem_or_io and devm_request_resource
to map the resource. I think this is the more "correct" way to do things
here, since we use the pdev resource, instead of creating a new one.
It's still a bit tricky, since we need to ensure that the resource is a
child of the fman region when it gets requested.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Acked-by: Camelia Groza <camelia.groza@nxp.com>
---

(no changes since v4)

Changes in v4:
- tricy -> tricky

Changes in v2:
- Fix warning if sizeof(void *) != sizeof(resource_size_t)

 .../net/ethernet/freescale/dpaa/dpaa_eth.c    |  4 +--
 .../ethernet/freescale/dpaa/dpaa_eth_sysfs.c  |  2 +-
 drivers/net/ethernet/freescale/fman/mac.c     | 35 +++++++------------
 drivers/net/ethernet/freescale/fman/mac.h     |  3 +-
 4 files changed, 17 insertions(+), 27 deletions(-)

Comments

Geert Uytterhoeven Oct. 17, 2022, 3:15 p.m. UTC | #1
Hi Sean,

On Sat, Sep 3, 2022 at 12:00 AM Sean Anderson <sean.anderson@seco.com> wrote:
> We don't need to remap the base address from the resource twice (once in
> mac_probe() and again in set_fman_mac_params()). We still need the
> resource to get the end address, but we can use a single function call
> to get both at once.
>
> While we're at it, use platform_get_mem_or_io and devm_request_resource
> to map the resource. I think this is the more "correct" way to do things
> here, since we use the pdev resource, instead of creating a new one.
> It's still a bit tricky, since we need to ensure that the resource is a
> child of the fman region when it gets requested.
>
> Signed-off-by: Sean Anderson <sean.anderson@seco.com>
> Acked-by: Camelia Groza <camelia.groza@nxp.com>

Thanks for your patch, which is now commit 262f2b782e255b79
("net: fman: Map the base address once") in v6.1-rc1.

> --- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c
> +++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c
> @@ -18,7 +18,7 @@ static ssize_t dpaa_eth_show_addr(struct device *dev,
>
>         if (mac_dev)
>                 return sprintf(buf, "%llx",
> -                               (unsigned long long)mac_dev->res->start);
> +                               (unsigned long long)mac_dev->vaddr);

On 32-bit:

    warning: cast from pointer to integer of different size
[-Wpointer-to-int-cast]

Obviously you should cast to "uintptr_t" or "unsigned long" instead,
and change the "%llx" to "%p" or "%lx"...

However, taking a closer look:
  1. The old code exposed a physical address to user space, the new
     code exposes the mapped virtual address.
     Is that change intentional?
  2. Virtual addresses are useless in user space.
     Moreover, addresses printed by "%p" are obfuscated, as this is
     considered a security issue. Likewise for working around this by
     casting to an integer.

What's the real purpose of dpaa_eth_show_addr()?
Perhaps it should be removed?

>         else
>                 return sprintf(buf, "none");
>  }

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
Sean Anderson Oct. 17, 2022, 3:34 p.m. UTC | #2
On 10/17/22 11:15 AM, Geert Uytterhoeven wrote:
> Hi Sean,
> 
> On Sat, Sep 3, 2022 at 12:00 AM Sean Anderson <sean.anderson@seco.com> wrote:
>> We don't need to remap the base address from the resource twice (once in
>> mac_probe() and again in set_fman_mac_params()). We still need the
>> resource to get the end address, but we can use a single function call
>> to get both at once.
>>
>> While we're at it, use platform_get_mem_or_io and devm_request_resource
>> to map the resource. I think this is the more "correct" way to do things
>> here, since we use the pdev resource, instead of creating a new one.
>> It's still a bit tricky, since we need to ensure that the resource is a
>> child of the fman region when it gets requested.
>>
>> Signed-off-by: Sean Anderson <sean.anderson@seco.com>
>> Acked-by: Camelia Groza <camelia.groza@nxp.com>
> 
> Thanks for your patch, which is now commit 262f2b782e255b79
> ("net: fman: Map the base address once") in v6.1-rc1.
> 
>> --- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c
>> +++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c
>> @@ -18,7 +18,7 @@ static ssize_t dpaa_eth_show_addr(struct device *dev,
>>
>>         if (mac_dev)
>>                 return sprintf(buf, "%llx",
>> -                               (unsigned long long)mac_dev->res->start);
>> +                               (unsigned long long)mac_dev->vaddr);
> 
> On 32-bit:
> 
>     warning: cast from pointer to integer of different size
> [-Wpointer-to-int-cast]
> 
> Obviously you should cast to "uintptr_t" or "unsigned long" instead,
> and change the "%llx" to "%p" or "%lx"...

Isn't there a %px for this purpose?

> However, taking a closer look:
>   1. The old code exposed a physical address to user space, the new
>      code exposes the mapped virtual address.
>      Is that change intentional?

No, this is not intentional. So to make this backwards-compatible, I
suppose I need a virt_to_phys?

>   2. Virtual addresses are useless in user space.
>      Moreover, addresses printed by "%p" are obfuscated, as this is
>      considered a security issue. Likewise for working around this by
>      casting to an integer.

Yes, you're right that this probably shouldn't be exposed to userspace.

> What's the real purpose of dpaa_eth_show_addr()?

I have no idea. This is a question for Madalin.

> Perhaps it should be removed?

That would be reasonable IMO.

--Sean

>>         else
>>                 return sprintf(buf, "none");
>>  }
> 
> Gr{oetje,eeting}s,
> 
>                         Geert
> 
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
> 
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like that.
>                                 -- Linus Torvalds
>
Geert Uytterhoeven Oct. 17, 2022, 3:49 p.m. UTC | #3
Hi Sean,

On Mon, Oct 17, 2022 at 5:34 PM Sean Anderson <sean.anderson@seco.com> wrote:
> On 10/17/22 11:15 AM, Geert Uytterhoeven wrote:
> > On Sat, Sep 3, 2022 at 12:00 AM Sean Anderson <sean.anderson@seco.com> wrote:
> >> We don't need to remap the base address from the resource twice (once in
> >> mac_probe() and again in set_fman_mac_params()). We still need the
> >> resource to get the end address, but we can use a single function call
> >> to get both at once.
> >>
> >> While we're at it, use platform_get_mem_or_io and devm_request_resource
> >> to map the resource. I think this is the more "correct" way to do things
> >> here, since we use the pdev resource, instead of creating a new one.
> >> It's still a bit tricky, since we need to ensure that the resource is a
> >> child of the fman region when it gets requested.
> >>
> >> Signed-off-by: Sean Anderson <sean.anderson@seco.com>
> >> Acked-by: Camelia Groza <camelia.groza@nxp.com>
> >
> > Thanks for your patch, which is now commit 262f2b782e255b79
> > ("net: fman: Map the base address once") in v6.1-rc1.
> >
> >> --- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c
> >> +++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c
> >> @@ -18,7 +18,7 @@ static ssize_t dpaa_eth_show_addr(struct device *dev,
> >>
> >>         if (mac_dev)
> >>                 return sprintf(buf, "%llx",
> >> -                               (unsigned long long)mac_dev->res->start);
> >> +                               (unsigned long long)mac_dev->vaddr);
> >
> > On 32-bit:
> >
> >     warning: cast from pointer to integer of different size
> > [-Wpointer-to-int-cast]
> >
> > Obviously you should cast to "uintptr_t" or "unsigned long" instead,
> > and change the "%llx" to "%p" or "%lx"...
>
> Isn't there a %px for this purpose?

Yes there is.  But if it makes sense to use that depends on the
still to be answered questions at the bottom...

> > However, taking a closer look:
> >   1. The old code exposed a physical address to user space, the new
> >      code exposes the mapped virtual address.
> >      Is that change intentional?
>
> No, this is not intentional. So to make this backwards-compatible, I
> suppose I need a virt_to_phys?

I think virt_to_phys() will work only on real memory, not on MMIO,
so you may need to reintroduce the resource again.

> >   2. Virtual addresses are useless in user space.
> >      Moreover, addresses printed by "%p" are obfuscated, as this is
> >      considered a security issue. Likewise for working around this by
> >      casting to an integer.
>
> Yes, you're right that this probably shouldn't be exposed to userspace.
>
> > What's the real purpose of dpaa_eth_show_addr()?
>
> I have no idea. This is a question for Madalin.
>
> > Perhaps it should be removed?
>
> That would be reasonable IMO.

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
diff mbox series

Patch

diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
index e974d90f15e3..02b588c46fcf 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
@@ -218,8 +218,8 @@  static int dpaa_netdev_init(struct net_device *net_dev,
 	net_dev->netdev_ops = dpaa_ops;
 	mac_addr = priv->mac_dev->addr;
 
-	net_dev->mem_start = priv->mac_dev->res->start;
-	net_dev->mem_end = priv->mac_dev->res->end;
+	net_dev->mem_start = (unsigned long)priv->mac_dev->vaddr;
+	net_dev->mem_end = (unsigned long)priv->mac_dev->vaddr_end;
 
 	net_dev->min_mtu = ETH_MIN_MTU;
 	net_dev->max_mtu = dpaa_get_max_mtu();
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c b/drivers/net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c
index 4fee74c024bd..258eb6c8f4c0 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c
@@ -18,7 +18,7 @@  static ssize_t dpaa_eth_show_addr(struct device *dev,
 
 	if (mac_dev)
 		return sprintf(buf, "%llx",
-				(unsigned long long)mac_dev->res->start);
+				(unsigned long long)mac_dev->vaddr);
 	else
 		return sprintf(buf, "none");
 }
diff --git a/drivers/net/ethernet/freescale/fman/mac.c b/drivers/net/ethernet/freescale/fman/mac.c
index 7afedd4995c9..62af81c0c942 100644
--- a/drivers/net/ethernet/freescale/fman/mac.c
+++ b/drivers/net/ethernet/freescale/fman/mac.c
@@ -28,7 +28,6 @@  MODULE_LICENSE("Dual BSD/GPL");
 MODULE_DESCRIPTION("FSL FMan MAC API based driver");
 
 struct mac_priv_s {
-	void __iomem			*vaddr;
 	u8				cell_index;
 	struct fman			*fman;
 	/* List of multicast addresses */
@@ -63,12 +62,7 @@  int set_fman_mac_params(struct mac_device *mac_dev,
 {
 	struct mac_priv_s *priv = mac_dev->priv;
 
-	params->base_addr = (typeof(params->base_addr))
-		devm_ioremap(mac_dev->dev, mac_dev->res->start,
-			     resource_size(mac_dev->res));
-	if (!params->base_addr)
-		return -ENOMEM;
-
+	params->base_addr = mac_dev->vaddr;
 	memcpy(&params->addr, mac_dev->addr, sizeof(mac_dev->addr));
 	params->max_speed	= priv->max_speed;
 	params->phy_if		= mac_dev->phy_if;
@@ -305,7 +299,7 @@  static int mac_probe(struct platform_device *_of_dev)
 	struct device_node	*mac_node, *dev_node;
 	struct mac_device	*mac_dev;
 	struct platform_device	*of_dev;
-	struct resource		 res;
+	struct resource		*res;
 	struct mac_priv_s	*priv;
 	u32			 val;
 	u8			fman_id;
@@ -368,30 +362,25 @@  static int mac_probe(struct platform_device *_of_dev)
 	of_node_put(dev_node);
 
 	/* Get the address of the memory mapped registers */
-	err = of_address_to_resource(mac_node, 0, &res);
-	if (err < 0) {
-		dev_err(dev, "of_address_to_resource(%pOF) = %d\n",
-			mac_node, err);
-		goto _return_of_node_put;
+	res = platform_get_mem_or_io(_of_dev, 0);
+	if (!res) {
+		dev_err(dev, "could not get registers\n");
+		return -EINVAL;
 	}
 
-	mac_dev->res = __devm_request_region(dev,
-					     fman_get_mem_region(priv->fman),
-					     res.start, resource_size(&res),
-					     "mac");
-	if (!mac_dev->res) {
-		dev_err(dev, "__devm_request_mem_region(mac) failed\n");
-		err = -EBUSY;
+	err = devm_request_resource(dev, fman_get_mem_region(priv->fman), res);
+	if (err) {
+		dev_err_probe(dev, err, "could not request resource\n");
 		goto _return_of_node_put;
 	}
 
-	priv->vaddr = devm_ioremap(dev, mac_dev->res->start,
-				   resource_size(mac_dev->res));
-	if (!priv->vaddr) {
+	mac_dev->vaddr = devm_ioremap(dev, res->start, resource_size(res));
+	if (!mac_dev->vaddr) {
 		dev_err(dev, "devm_ioremap() failed\n");
 		err = -EIO;
 		goto _return_of_node_put;
 	}
+	mac_dev->vaddr_end = mac_dev->vaddr + resource_size(res);
 
 	if (!of_device_is_available(mac_node)) {
 		err = -ENODEV;
diff --git a/drivers/net/ethernet/freescale/fman/mac.h b/drivers/net/ethernet/freescale/fman/mac.h
index da410a7d00c9..7aa71b05bd3e 100644
--- a/drivers/net/ethernet/freescale/fman/mac.h
+++ b/drivers/net/ethernet/freescale/fman/mac.h
@@ -19,8 +19,9 @@  struct fman_mac;
 struct mac_priv_s;
 
 struct mac_device {
+	void __iomem		*vaddr;
+	void __iomem		*vaddr_end;
 	struct device		*dev;
-	struct resource		*res;
 	u8			 addr[ETH_ALEN];
 	struct fman_port	*port[2];
 	u32			 if_support;