| Message ID | 1380826568-18587-1-git-send-email-sebastian.hesselbarth@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Thu, Oct 03, 2013 at 08:56:08PM +0200, Sebastian Hesselbarth wrote: > Ethernet IP on Kirkwood SoCs loose their MAC address register content > if clock gated. To allow modular ethernet driver setups and gated clocks > also on non-DT capable bootloaders, we fixup port device nodes with no > valid MAC address property. This patch copies MAC address register > contents set up by bootloaders early, notably before ethernet clocks > are gated. While at it, also reorder call sequence in _dt_init. > > Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com> > --- > Changelog: > v1->v2: > - check for gated clock before accessing eth registers > (Suggested by Andrew Lunn) > > TBH, as there is no "official" clk_is_enabled but just the __clk_is_enabled > one, this is close to abuse of CCF API. Actually, there should either be an > clk_is_enabled or we enable the clock, copy a possible invalid MAC address > and disable the clock again. (Added Mike, who might have an opinion about > clk_is_enabled) > > Cc: Jason Cooper <jason@lakedaemon.net> > Cc: Andrew Lunn <andrew@lunn.ch> > Cc: Russell King <linux@arm.linux.org.uk> > Cc: Grant Likely <grant.likely@linaro.org> > Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> > Cc: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> > Cc: Ezequiel Garcia <ezequiel.garcia@free-electrons.com> > Cc: Mike Turquette <mturquette@linaro.org> > Cc: linux-arm-kernel@lists.infradead.org > Cc: linux-kernel@vger.kernel.org > --- > arch/arm/mach-kirkwood/board-dt.c | 77 +++++++++++++++++++++++++++++++++++-- > 1 file changed, 74 insertions(+), 3 deletions(-) > > diff --git a/arch/arm/mach-kirkwood/board-dt.c b/arch/arm/mach-kirkwood/board-dt.c > index 82d3ad8..f5f35bb 100644 > --- a/arch/arm/mach-kirkwood/board-dt.c > +++ b/arch/arm/mach-kirkwood/board-dt.c > @@ -13,6 +13,8 @@ > #include <linux/kernel.h> > #include <linux/init.h> > #include <linux/of.h> > +#include <linux/of_address.h> > +#include <linux/of_net.h> > #include <linux/of_platform.h> > #include <linux/clk-provider.h> > #include <linux/clocksource.h> > @@ -66,6 +68,76 @@ static void __init kirkwood_legacy_clk_init(void) > clk_prepare_enable(clk); > } > > +#define MV643XX_ETH_MAC_ADDR_LOW 0x0414 > +#define MV643XX_ETH_MAC_ADDR_HIGH 0x0418 > + > +static void __init kirkwood_dt_eth_fixup(void) > +{ > + struct device_node *np; > + > + /* > + * The ethernet interfaces forget the MAC address assigned by u-boot > + * if the clocks are turned off. Usually, u-boot on kirkwood boards > + * has no DT support to properly set local-mac-address property. > + * As a workaround, we get the MAC address from mv643xx_eth registers > + * and update the port device node if no valid MAC address is set. > + */ > + for_each_compatible_node(np, NULL, "marvell,kirkwood-eth-port") { > + struct device_node *pnp = of_get_parent(np); > + struct clk *clk; > + struct property *pmac; > + void __iomem *io; > + u8 *macaddr; > + u32 reg; > + > + if (!pnp || !of_device_is_available(pnp)) > + continue; > + > + clk = of_clk_get(pnp, 0); > + if (!__clk_is_enabled(clk) || of_get_mac_address(np)) { Mm.. just as you say this usage of __clk_is_enabled is awkard to say the least. I'm wondering: is the clock really disabled if the device is not available (i.e. status == 'ok')? In other words: isn't the !of_device_is_available() test enough? Or maybe I'm missing something?
On Thu, Oct 03, 2013 at 04:02:02PM -0300, Ezequiel Garcia wrote: > On Thu, Oct 03, 2013 at 08:56:08PM +0200, Sebastian Hesselbarth wrote: > > Ethernet IP on Kirkwood SoCs loose their MAC address register content > > if clock gated. To allow modular ethernet driver setups and gated clocks > > also on non-DT capable bootloaders, we fixup port device nodes with no > > valid MAC address property. This patch copies MAC address register > > contents set up by bootloaders early, notably before ethernet clocks > > are gated. While at it, also reorder call sequence in _dt_init. > > > > Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com> > > --- > > Changelog: > > v1->v2: > > - check for gated clock before accessing eth registers > > (Suggested by Andrew Lunn) > > > > TBH, as there is no "official" clk_is_enabled but just the __clk_is_enabled > > one, this is close to abuse of CCF API. Actually, there should either be an > > clk_is_enabled or we enable the clock, copy a possible invalid MAC address > > and disable the clock again. (Added Mike, who might have an opinion about > > clk_is_enabled) > > > > Cc: Jason Cooper <jason@lakedaemon.net> > > Cc: Andrew Lunn <andrew@lunn.ch> > > Cc: Russell King <linux@arm.linux.org.uk> > > Cc: Grant Likely <grant.likely@linaro.org> > > Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> > > Cc: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> > > Cc: Ezequiel Garcia <ezequiel.garcia@free-electrons.com> > > Cc: Mike Turquette <mturquette@linaro.org> > > Cc: linux-arm-kernel@lists.infradead.org > > Cc: linux-kernel@vger.kernel.org > > --- > > arch/arm/mach-kirkwood/board-dt.c | 77 +++++++++++++++++++++++++++++++++++-- > > 1 file changed, 74 insertions(+), 3 deletions(-) > > > > diff --git a/arch/arm/mach-kirkwood/board-dt.c b/arch/arm/mach-kirkwood/board-dt.c > > index 82d3ad8..f5f35bb 100644 > > --- a/arch/arm/mach-kirkwood/board-dt.c > > +++ b/arch/arm/mach-kirkwood/board-dt.c > > @@ -13,6 +13,8 @@ > > #include <linux/kernel.h> > > #include <linux/init.h> > > #include <linux/of.h> > > +#include <linux/of_address.h> > > +#include <linux/of_net.h> > > #include <linux/of_platform.h> > > #include <linux/clk-provider.h> > > #include <linux/clocksource.h> > > @@ -66,6 +68,76 @@ static void __init kirkwood_legacy_clk_init(void) > > clk_prepare_enable(clk); > > } > > > > +#define MV643XX_ETH_MAC_ADDR_LOW 0x0414 > > +#define MV643XX_ETH_MAC_ADDR_HIGH 0x0418 > > + > > +static void __init kirkwood_dt_eth_fixup(void) > > +{ > > + struct device_node *np; > > + > > + /* > > + * The ethernet interfaces forget the MAC address assigned by u-boot > > + * if the clocks are turned off. Usually, u-boot on kirkwood boards > > + * has no DT support to properly set local-mac-address property. > > + * As a workaround, we get the MAC address from mv643xx_eth registers > > + * and update the port device node if no valid MAC address is set. > > + */ > > + for_each_compatible_node(np, NULL, "marvell,kirkwood-eth-port") { > > + struct device_node *pnp = of_get_parent(np); > > + struct clk *clk; > > + struct property *pmac; > > + void __iomem *io; > > + u8 *macaddr; > > + u32 reg; > > + > > + if (!pnp || !of_device_is_available(pnp)) > > + continue; > > + > > + clk = of_clk_get(pnp, 0); > > + if (!__clk_is_enabled(clk) || of_get_mac_address(np)) { > > Mm.. just as you say this usage of __clk_is_enabled is awkard to say the > least. > > I'm wondering: is the clock really disabled if the device is not > available (i.e. status == 'ok')? In other words: isn't the > !of_device_is_available() test enough? Well, this stemmed from JasonG's scenario where the second iface is shut off by the bootloader. Although one could argue that the bootloader should then update the dtb to mark that node as disabled... thx, Jason.
On Thu, Oct 03, 2013 at 03:04:37PM -0400, Jason Cooper wrote: > > I'm wondering: is the clock really disabled if the device is not > > available (i.e. status == 'ok')? In other words: isn't the > > !of_device_is_available() test enough? > > Well, this stemmed from JasonG's scenario where the second iface is > shut off by the bootloader. Although one could argue that the > bootloader should then update the dtb to mark that node as disabled... Right, that is what we do here, only the first eth is present in the dt, the second is gated and powered down (noting that Linux doesn't know how to power it up :() This check is only to prevent CPU lockup if the firmware has included a DT node for ethernet, not included the MAC address and turned off the clock. Sebastian, does __clk_enabled work properly for the mvebu clock provider? I don't see a clk_ops.is_enabled for mvebu.. (don't know much about clk) Also, I think you should move the pr_err above the clock test, or make a special pr_err for the clock test as well. Having the clock gated, no mac address and a dt node is still a fw bug. Regards, Jason
On 10/03/2013 09:44 PM, Jason Gunthorpe wrote: > On Thu, Oct 03, 2013 at 03:04:37PM -0400, Jason Cooper wrote: > >>> I'm wondering: is the clock really disabled if the device is not >>> available (i.e. status == 'ok')? In other words: isn't the >>> !of_device_is_available() test enough? >> >> Well, this stemmed from JasonG's scenario where the second iface is >> shut off by the bootloader. Although one could argue that the >> bootloader should then update the dtb to mark that node as disabled... > > Right, that is what we do here, only the first eth is present in the > dt, the second is gated and powered down (noting that Linux doesn't > know how to power it up :() > > This check is only to prevent CPU lockup if the firmware has included > a DT node for ethernet, not included the MAC address and turned off > the clock. > > Sebastian, does __clk_enabled work properly for the mvebu clock > provider? I don't see a clk_ops.is_enabled for mvebu.. (don't know > much about clk) (joint answer for all three above) A node with status != "okay" means "hardware/board does not allow you to use that port at all, i.e. no ethernet jack"; disabling the clock just means "I don't expect _all_ users want to use this jack, but it is connected". So there _is_ a difference here. clk-gate, which is the underlying clk provider for our gates, does provide .is_enabled and reads the register. I haven't tested this, due to no KW hardware available right now, but it should test for disabled clocks as expected. > Also, I think you should move the pr_err above the clock test, or make > a special pr_err for the clock test as well. Having the clock gated, > no mac address and a dt node is still a fw bug. Well, we have 4 cases here: - status != "okay": skip - of_get_mac_address() != NULL: skip (no registers read) - !__clk_is_enabled(): skip and warn with FW_BUG about disabled clocks but no valid MAC address set. - else, read registers, store MAC in DT, warn with FW_BUG about enabled clocks and no valid MAC address set. Leaves missing public clk_is_enabled(), which I can provide a patch for if Mike agrees (and I haven't missed anything without really looking into CCF). Sebastian
> Sebastian, does __clk_enabled work properly for the mvebu clock > provider? I don't see a clk_ops.is_enabled for mvebu.. (don't know > much about clk) Hi Jason It is implemented in drivers/clk/clk-gate.c, which is what mvebu is using. Andrew
diff --git a/arch/arm/mach-kirkwood/board-dt.c b/arch/arm/mach-kirkwood/board-dt.c index 82d3ad8..f5f35bb 100644 --- a/arch/arm/mach-kirkwood/board-dt.c +++ b/arch/arm/mach-kirkwood/board-dt.c @@ -13,6 +13,8 @@ #include <linux/kernel.h> #include <linux/init.h> #include <linux/of.h> +#include <linux/of_address.h> +#include <linux/of_net.h> #include <linux/of_platform.h> #include <linux/clk-provider.h> #include <linux/clocksource.h> @@ -66,6 +68,76 @@ static void __init kirkwood_legacy_clk_init(void) clk_prepare_enable(clk); } +#define MV643XX_ETH_MAC_ADDR_LOW 0x0414 +#define MV643XX_ETH_MAC_ADDR_HIGH 0x0418 + +static void __init kirkwood_dt_eth_fixup(void) +{ + struct device_node *np; + + /* + * The ethernet interfaces forget the MAC address assigned by u-boot + * if the clocks are turned off. Usually, u-boot on kirkwood boards + * has no DT support to properly set local-mac-address property. + * As a workaround, we get the MAC address from mv643xx_eth registers + * and update the port device node if no valid MAC address is set. + */ + for_each_compatible_node(np, NULL, "marvell,kirkwood-eth-port") { + struct device_node *pnp = of_get_parent(np); + struct clk *clk; + struct property *pmac; + void __iomem *io; + u8 *macaddr; + u32 reg; + + if (!pnp || !of_device_is_available(pnp)) + continue; + + clk = of_clk_get(pnp, 0); + if (!__clk_is_enabled(clk) || of_get_mac_address(np)) { + clk_put(clk); + continue; + } + clk_put(clk); + + pr_err(FW_BUG "%s: local-mac-address is not set\n", + np->full_name); + + io = of_iomap(pnp, 0); + if (!io) + continue; + + pmac = kzalloc(sizeof(*pmac) + 6, GFP_KERNEL); + if (!pmac) { + iounmap(io); + continue; + } + + pmac->value = pmac + 1; + pmac->length = 6; + pmac->name = kstrdup("local-mac-address", GFP_KERNEL); + if (!pmac->name) { + kfree(pmac); + iounmap(io); + continue; + } + + macaddr = pmac->value; + reg = readl(io + MV643XX_ETH_MAC_ADDR_HIGH); + macaddr[0] = (reg >> 24) & 0xff; + macaddr[1] = (reg >> 16) & 0xff; + macaddr[2] = (reg >> 8) & 0xff; + macaddr[3] = reg & 0xff; + + reg = readl(io + MV643XX_ETH_MAC_ADDR_LOW); + macaddr[4] = (reg >> 8) & 0xff; + macaddr[5] = reg & 0xff; + + of_update_property(np, pmac); + iounmap(io); + } +} + static void __init kirkwood_dt_time_init(void) { of_clk_init(NULL); @@ -97,11 +169,10 @@ static void __init kirkwood_dt_init(void) kirkwood_l2_init(); kirkwood_cpufreq_init(); - + kirkwood_cpuidle_init(); /* Setup clocks for legacy devices */ kirkwood_legacy_clk_init(); - - kirkwood_cpuidle_init(); + kirkwood_dt_eth_fixup(); #ifdef CONFIG_KEXEC kexec_reinit = kirkwood_enable_pcie;
Ethernet IP on Kirkwood SoCs loose their MAC address register content if clock gated. To allow modular ethernet driver setups and gated clocks also on non-DT capable bootloaders, we fixup port device nodes with no valid MAC address property. This patch copies MAC address register contents set up by bootloaders early, notably before ethernet clocks are gated. While at it, also reorder call sequence in _dt_init. Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com> --- Changelog: v1->v2: - check for gated clock before accessing eth registers (Suggested by Andrew Lunn) TBH, as there is no "official" clk_is_enabled but just the __clk_is_enabled one, this is close to abuse of CCF API. Actually, there should either be an clk_is_enabled or we enable the clock, copy a possible invalid MAC address and disable the clock again. (Added Mike, who might have an opinion about clk_is_enabled) Cc: Jason Cooper <jason@lakedaemon.net> Cc: Andrew Lunn <andrew@lunn.ch> Cc: Russell King <linux@arm.linux.org.uk> Cc: Grant Likely <grant.likely@linaro.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Cc: Ezequiel Garcia <ezequiel.garcia@free-electrons.com> Cc: Mike Turquette <mturquette@linaro.org> Cc: linux-arm-kernel@lists.infradead.org Cc: linux-kernel@vger.kernel.org --- arch/arm/mach-kirkwood/board-dt.c | 77 +++++++++++++++++++++++++++++++++++-- 1 file changed, 74 insertions(+), 3 deletions(-)