Re: [PATCH v5 2/8] clk: rockchip: rk3036: fix and add node id for emac clock
Am Sonntag, 21. Februar 2016, 10:45:11 schrieb Caesar Wang: > Heiko, > > 在 2016年02月21日 10:26, Heiko Stuebner 写道: > > Hi Caesar, Xing, > > > > Am Dienstag, 2. Februar 2016, 11:48:19 schrieb Caesar Wang: > >> From: zhengxing> >> > >> In the emac driver, we need to refer HCLK_MAC since there are > >> only 3PLLs (APLL/GPLL/DPLL) on the rk3036, most clock are under the > >> GPLL, and it is unable to provide the accurate rate for mac_ref which > >> need to 50MHz probability, we should let it under the DPLL and are > >> able to set the freq which integer multiples of 50MHz, so we add these > >> emac node for reference. > >> > >> Signed-off-by: Xing Zheng > >> Signed-off-by: Caesar Wang > > > > [...] > > > >> --- a/drivers/clk/rockchip/clk-rk3036.c > >> +++ b/drivers/clk/rockchip/clk-rk3036.c > >> @@ -343,8 +343,11 @@ static struct rockchip_clk_branch > >> rk3036_clk_branches[] __initdata = { RK2928_CLKSEL_CON(16), 0, 2, > >> MFLAGS, > >> 2, 5, DFLAGS, > >> > >>RK2928_CLKGATE_CON(10), 5, GFLAGS), > >> > >> - COMPOSITE_NOGATE(0, "mac_pll_src", mux_pll_src_3plls_p, 0, > >> - RK2928_CLKSEL_CON(21), 0, 2, MFLAGS, 9, 5, DFLAGS), > >> + MUX(SCLK_MACPLL, "mac_pll_pre", mux_pll_src_3plls_p, 0, > >> + RK2928_CLKSEL_CON(21), 0, 2, MFLAGS), > >> + DIV(0, "mac_pll_src", "mac_pll_pre", 0, > >> + RK2928_CLKSEL_CON(21), 9, 5, DFLAGS), > >> + > > > > CLK_SET_RATE_NO_REPARENT should do the trick as well. > > > > And the whole hclk + clkid part should be separate patches. I took the > > liberty of splitting them already in [0] to see if I could get the emac > > running on my kylin board. > > > > Probing emac + phy does suceed, but there is no link-detection. > > Building your kylin-develop4.4 branch [1] results in the same (aka no > > transmission). > > > > Only with the original uboot + 4.1-based kernel that was already on the > > device did I manage to get a network connection. > > I guess you need apply the uboot patch[0]. > > patch[0]: > http://lists.denx.de/pipermail/u-boot/2016-February/245814.html > > or get the uboot from rockchip github: > https://github.com/rockchip-linux/u-boot/commits/rk3036 I have a uboot with the ddr-freq change already, and the clocks look like expected (both on the android system as well as on my mainline kernel with the adapted clock change): pll_dpll 11 4 0 0 dpll 11 4 0 0 mac_pll_src 115000 0 0 mac_clk_ref 335000 0 0 mac_clk 112500 0 0 The probing itself also seems to work like a charm: [3.064437] rockchip_emac 1020.ethernet: ARC EMAC detected with id: 0x7fd02 [3.072123] rockchip_emac 1020.ethernet: IRQ is 29 [3.090055] rockchip_emac 1020.ethernet: MAC address is now a2:d6:55:be:be:ad [3.100732] libphy: Synopsys MII Bus: probed [3.202432] rockchip_emac 1020.ethernet: connected to RTL8201F Ethernet phy with id 0x1cc816 (same with the real generic phy) Digging a bit into the phy status it looks like it never detects the link, so my current guess is that I have some sort of clock issue that isn't shown in the general clock summary, as the phy should only need the mac clock to do the link negotiation. I'll dig some more. Heiko > > Is there some additional setup missing somewhere? > > > > > > Heiko > > > > [0] https://github.com/mmind/linux-rockchip/commits/tmp/rk3036-emac > > The 3 additional patches are not strictly necessary there. > > > > [1] https://github.com/rockchip-linux/kernel/tree/kylin-develop4.4 > > The lastest kylin-develop.4.4.y from my github: > https://github.com/Caesar-github/rockchip/tree/kylin/develop-4.4.y > > > ___ > > Linux-rockchip mailing list > > linux-rockc...@lists.infradead.org > > http://lists.infradead.org/mailman/listinfo/linux-rockchip
Re: [PATCH v5 2/8] clk: rockchip: rk3036: fix and add node id for emac clock
Am Sonntag, 21. Februar 2016, 10:45:11 schrieb Caesar Wang: > Heiko, > > 在 2016年02月21日 10:26, Heiko Stuebner 写道: > > Hi Caesar, Xing, > > > > Am Dienstag, 2. Februar 2016, 11:48:19 schrieb Caesar Wang: > >> From: zhengxing > >> > >> In the emac driver, we need to refer HCLK_MAC since there are > >> only 3PLLs (APLL/GPLL/DPLL) on the rk3036, most clock are under the > >> GPLL, and it is unable to provide the accurate rate for mac_ref which > >> need to 50MHz probability, we should let it under the DPLL and are > >> able to set the freq which integer multiples of 50MHz, so we add these > >> emac node for reference. > >> > >> Signed-off-by: Xing Zheng > >> Signed-off-by: Caesar Wang > > > > [...] > > > >> --- a/drivers/clk/rockchip/clk-rk3036.c > >> +++ b/drivers/clk/rockchip/clk-rk3036.c > >> @@ -343,8 +343,11 @@ static struct rockchip_clk_branch > >> rk3036_clk_branches[] __initdata = { RK2928_CLKSEL_CON(16), 0, 2, > >> MFLAGS, > >> 2, 5, DFLAGS, > >> > >>RK2928_CLKGATE_CON(10), 5, GFLAGS), > >> > >> - COMPOSITE_NOGATE(0, "mac_pll_src", mux_pll_src_3plls_p, 0, > >> - RK2928_CLKSEL_CON(21), 0, 2, MFLAGS, 9, 5, DFLAGS), > >> + MUX(SCLK_MACPLL, "mac_pll_pre", mux_pll_src_3plls_p, 0, > >> + RK2928_CLKSEL_CON(21), 0, 2, MFLAGS), > >> + DIV(0, "mac_pll_src", "mac_pll_pre", 0, > >> + RK2928_CLKSEL_CON(21), 9, 5, DFLAGS), > >> + > > > > CLK_SET_RATE_NO_REPARENT should do the trick as well. > > > > And the whole hclk + clkid part should be separate patches. I took the > > liberty of splitting them already in [0] to see if I could get the emac > > running on my kylin board. > > > > Probing emac + phy does suceed, but there is no link-detection. > > Building your kylin-develop4.4 branch [1] results in the same (aka no > > transmission). > > > > Only with the original uboot + 4.1-based kernel that was already on the > > device did I manage to get a network connection. > > I guess you need apply the uboot patch[0]. > > patch[0]: > http://lists.denx.de/pipermail/u-boot/2016-February/245814.html > > or get the uboot from rockchip github: > https://github.com/rockchip-linux/u-boot/commits/rk3036 I have a uboot with the ddr-freq change already, and the clocks look like expected (both on the android system as well as on my mainline kernel with the adapted clock change): pll_dpll 11 4 0 0 dpll 11 4 0 0 mac_pll_src 115000 0 0 mac_clk_ref 335000 0 0 mac_clk 112500 0 0 The probing itself also seems to work like a charm: [3.064437] rockchip_emac 1020.ethernet: ARC EMAC detected with id: 0x7fd02 [3.072123] rockchip_emac 1020.ethernet: IRQ is 29 [3.090055] rockchip_emac 1020.ethernet: MAC address is now a2:d6:55:be:be:ad [3.100732] libphy: Synopsys MII Bus: probed [3.202432] rockchip_emac 1020.ethernet: connected to RTL8201F Ethernet phy with id 0x1cc816 (same with the real generic phy) Digging a bit into the phy status it looks like it never detects the link, so my current guess is that I have some sort of clock issue that isn't shown in the general clock summary, as the phy should only need the mac clock to do the link negotiation. I'll dig some more. Heiko > > Is there some additional setup missing somewhere? > > > > > > Heiko > > > > [0] https://github.com/mmind/linux-rockchip/commits/tmp/rk3036-emac > > The 3 additional patches are not strictly necessary there. > > > > [1] https://github.com/rockchip-linux/kernel/tree/kylin-develop4.4 > > The lastest kylin-develop.4.4.y from my github: > https://github.com/Caesar-github/rockchip/tree/kylin/develop-4.4.y > > > ___ > > Linux-rockchip mailing list > > linux-rockc...@lists.infradead.org > > http://lists.infradead.org/mailman/listinfo/linux-rockchip
Re: [PATCH v5 2/8] clk: rockchip: rk3036: fix and add node id for emac clock
Heiko, 在 2016年02月21日 10:26, Heiko Stuebner 写道: Hi Caesar, Xing, Am Dienstag, 2. Februar 2016, 11:48:19 schrieb Caesar Wang: From: zhengxingIn the emac driver, we need to refer HCLK_MAC since there are only 3PLLs (APLL/GPLL/DPLL) on the rk3036, most clock are under the GPLL, and it is unable to provide the accurate rate for mac_ref which need to 50MHz probability, we should let it under the DPLL and are able to set the freq which integer multiples of 50MHz, so we add these emac node for reference. Signed-off-by: Xing Zheng Signed-off-by: Caesar Wang [...] --- a/drivers/clk/rockchip/clk-rk3036.c +++ b/drivers/clk/rockchip/clk-rk3036.c @@ -343,8 +343,11 @@ static struct rockchip_clk_branch rk3036_clk_branches[] __initdata = { RK2928_CLKSEL_CON(16), 0, 2, MFLAGS, 2, 5, DFLAGS, RK2928_CLKGATE_CON(10), 5, GFLAGS), - COMPOSITE_NOGATE(0, "mac_pll_src", mux_pll_src_3plls_p, 0, - RK2928_CLKSEL_CON(21), 0, 2, MFLAGS, 9, 5, DFLAGS), + MUX(SCLK_MACPLL, "mac_pll_pre", mux_pll_src_3plls_p, 0, + RK2928_CLKSEL_CON(21), 0, 2, MFLAGS), + DIV(0, "mac_pll_src", "mac_pll_pre", 0, + RK2928_CLKSEL_CON(21), 9, 5, DFLAGS), + CLK_SET_RATE_NO_REPARENT should do the trick as well. And the whole hclk + clkid part should be separate patches. I took the liberty of splitting them already in [0] to see if I could get the emac running on my kylin board. Probing emac + phy does suceed, but there is no link-detection. Building your kylin-develop4.4 branch [1] results in the same (aka no transmission). Only with the original uboot + 4.1-based kernel that was already on the device did I manage to get a network connection. I guess you need apply the uboot patch[0]. patch[0]: http://lists.denx.de/pipermail/u-boot/2016-February/245814.html or get the uboot from rockchip github: https://github.com/rockchip-linux/u-boot/commits/rk3036 Is there some additional setup missing somewhere? Heiko [0] https://github.com/mmind/linux-rockchip/commits/tmp/rk3036-emac The 3 additional patches are not strictly necessary there. [1] https://github.com/rockchip-linux/kernel/tree/kylin-develop4.4 The lastest kylin-develop.4.4.y from my github: https://github.com/Caesar-github/rockchip/tree/kylin/develop-4.4.y ___ Linux-rockchip mailing list linux-rockc...@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-rockchip -- Thanks, Caesar
Re: [PATCH v5 2/8] clk: rockchip: rk3036: fix and add node id for emac clock
Heiko, 在 2016年02月21日 10:26, Heiko Stuebner 写道: Hi Caesar, Xing, Am Dienstag, 2. Februar 2016, 11:48:19 schrieb Caesar Wang: From: zhengxing In the emac driver, we need to refer HCLK_MAC since there are only 3PLLs (APLL/GPLL/DPLL) on the rk3036, most clock are under the GPLL, and it is unable to provide the accurate rate for mac_ref which need to 50MHz probability, we should let it under the DPLL and are able to set the freq which integer multiples of 50MHz, so we add these emac node for reference. Signed-off-by: Xing Zheng Signed-off-by: Caesar Wang [...] --- a/drivers/clk/rockchip/clk-rk3036.c +++ b/drivers/clk/rockchip/clk-rk3036.c @@ -343,8 +343,11 @@ static struct rockchip_clk_branch rk3036_clk_branches[] __initdata = { RK2928_CLKSEL_CON(16), 0, 2, MFLAGS, 2, 5, DFLAGS, RK2928_CLKGATE_CON(10), 5, GFLAGS), - COMPOSITE_NOGATE(0, "mac_pll_src", mux_pll_src_3plls_p, 0, - RK2928_CLKSEL_CON(21), 0, 2, MFLAGS, 9, 5, DFLAGS), + MUX(SCLK_MACPLL, "mac_pll_pre", mux_pll_src_3plls_p, 0, + RK2928_CLKSEL_CON(21), 0, 2, MFLAGS), + DIV(0, "mac_pll_src", "mac_pll_pre", 0, + RK2928_CLKSEL_CON(21), 9, 5, DFLAGS), + CLK_SET_RATE_NO_REPARENT should do the trick as well. And the whole hclk + clkid part should be separate patches. I took the liberty of splitting them already in [0] to see if I could get the emac running on my kylin board. Probing emac + phy does suceed, but there is no link-detection. Building your kylin-develop4.4 branch [1] results in the same (aka no transmission). Only with the original uboot + 4.1-based kernel that was already on the device did I manage to get a network connection. I guess you need apply the uboot patch[0]. patch[0]: http://lists.denx.de/pipermail/u-boot/2016-February/245814.html or get the uboot from rockchip github: https://github.com/rockchip-linux/u-boot/commits/rk3036 Is there some additional setup missing somewhere? Heiko [0] https://github.com/mmind/linux-rockchip/commits/tmp/rk3036-emac The 3 additional patches are not strictly necessary there. [1] https://github.com/rockchip-linux/kernel/tree/kylin-develop4.4 The lastest kylin-develop.4.4.y from my github: https://github.com/Caesar-github/rockchip/tree/kylin/develop-4.4.y ___ Linux-rockchip mailing list linux-rockc...@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-rockchip -- Thanks, Caesar
Re: [PATCH v5 2/8] clk: rockchip: rk3036: fix and add node id for emac clock
Hi Caesar, Xing, Am Dienstag, 2. Februar 2016, 11:48:19 schrieb Caesar Wang: > From: zhengxing> > In the emac driver, we need to refer HCLK_MAC since there are > only 3PLLs (APLL/GPLL/DPLL) on the rk3036, most clock are under the > GPLL, and it is unable to provide the accurate rate for mac_ref which > need to 50MHz probability, we should let it under the DPLL and are > able to set the freq which integer multiples of 50MHz, so we add these > emac node for reference. > > Signed-off-by: Xing Zheng > Signed-off-by: Caesar Wang [...] > --- a/drivers/clk/rockchip/clk-rk3036.c > +++ b/drivers/clk/rockchip/clk-rk3036.c > @@ -343,8 +343,11 @@ static struct rockchip_clk_branch > rk3036_clk_branches[] __initdata = { RK2928_CLKSEL_CON(16), 0, 2, MFLAGS, > 2, 5, DFLAGS, > RK2928_CLKGATE_CON(10), 5, GFLAGS), > > - COMPOSITE_NOGATE(0, "mac_pll_src", mux_pll_src_3plls_p, 0, > - RK2928_CLKSEL_CON(21), 0, 2, MFLAGS, 9, 5, DFLAGS), > + MUX(SCLK_MACPLL, "mac_pll_pre", mux_pll_src_3plls_p, 0, > + RK2928_CLKSEL_CON(21), 0, 2, MFLAGS), > + DIV(0, "mac_pll_src", "mac_pll_pre", 0, > + RK2928_CLKSEL_CON(21), 9, 5, DFLAGS), > + CLK_SET_RATE_NO_REPARENT should do the trick as well. And the whole hclk + clkid part should be separate patches. I took the liberty of splitting them already in [0] to see if I could get the emac running on my kylin board. Probing emac + phy does suceed, but there is no link-detection. Building your kylin-develop4.4 branch [1] results in the same (aka no transmission). Only with the original uboot + 4.1-based kernel that was already on the device did I manage to get a network connection. Is there some additional setup missing somewhere? Heiko [0] https://github.com/mmind/linux-rockchip/commits/tmp/rk3036-emac The 3 additional patches are not strictly necessary there. [1] https://github.com/rockchip-linux/kernel/tree/kylin-develop4.4
Re: [PATCH v5 2/8] clk: rockchip: rk3036: fix and add node id for emac clock
Hi Caesar, Xing, Am Dienstag, 2. Februar 2016, 11:48:19 schrieb Caesar Wang: > From: zhengxing > > In the emac driver, we need to refer HCLK_MAC since there are > only 3PLLs (APLL/GPLL/DPLL) on the rk3036, most clock are under the > GPLL, and it is unable to provide the accurate rate for mac_ref which > need to 50MHz probability, we should let it under the DPLL and are > able to set the freq which integer multiples of 50MHz, so we add these > emac node for reference. > > Signed-off-by: Xing Zheng > Signed-off-by: Caesar Wang [...] > --- a/drivers/clk/rockchip/clk-rk3036.c > +++ b/drivers/clk/rockchip/clk-rk3036.c > @@ -343,8 +343,11 @@ static struct rockchip_clk_branch > rk3036_clk_branches[] __initdata = { RK2928_CLKSEL_CON(16), 0, 2, MFLAGS, > 2, 5, DFLAGS, > RK2928_CLKGATE_CON(10), 5, GFLAGS), > > - COMPOSITE_NOGATE(0, "mac_pll_src", mux_pll_src_3plls_p, 0, > - RK2928_CLKSEL_CON(21), 0, 2, MFLAGS, 9, 5, DFLAGS), > + MUX(SCLK_MACPLL, "mac_pll_pre", mux_pll_src_3plls_p, 0, > + RK2928_CLKSEL_CON(21), 0, 2, MFLAGS), > + DIV(0, "mac_pll_src", "mac_pll_pre", 0, > + RK2928_CLKSEL_CON(21), 9, 5, DFLAGS), > + CLK_SET_RATE_NO_REPARENT should do the trick as well. And the whole hclk + clkid part should be separate patches. I took the liberty of splitting them already in [0] to see if I could get the emac running on my kylin board. Probing emac + phy does suceed, but there is no link-detection. Building your kylin-develop4.4 branch [1] results in the same (aka no transmission). Only with the original uboot + 4.1-based kernel that was already on the device did I manage to get a network connection. Is there some additional setup missing somewhere? Heiko [0] https://github.com/mmind/linux-rockchip/commits/tmp/rk3036-emac The 3 additional patches are not strictly necessary there. [1] https://github.com/rockchip-linux/kernel/tree/kylin-develop4.4