From: Simon Goldschmidt Date: Wed, 20 Nov 2019 20:49:51 +0100 Subject: [U-Boot] [PATCH v3] spi: cadence_qspi: support DM_CLK In-Reply-To: References: <20191111214251.4112-1-simon.k.r.goldschmidt@gmail.com> <9423d84d-e25f-65b3-3281-009860f7206a@ti.com> <61139e9b-9e21-2d5e-f75f-b0c7d53e58d1@ti.com> Message-ID: List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: u-boot@lists.denx.de Marek, Am 12.11.2019 um 12:57 schrieb Simon Goldschmidt: > On Tue, Nov 12, 2019 at 12:40 PM Vignesh Raghavendra wrote: >> >> >> >> On 12/11/19 4:57 PM, Simon Goldschmidt wrote: >>> On Tue, Nov 12, 2019 at 10:30 AM Tan, Ley Foon wrote: >>>> >> [...] >>>>>> But, unfortunately, such stub does not exists for clk_get_rate(). >>>>>> So on platforms w/o CONFIG_CLK set: >>>>>> >>>>>> arm-linux-gnueabihf-ld.bfd: drivers/spi/built-in.o: in function >>>>> `cadence_spi_probe': >>>>>> /home/a0132425/workspace/u-boot/drivers/spi/cadence_qspi.c:184: >>>>> undefined reference to `clk_get_rate' >>>>>> Makefile:1647: recipe for target 'u-boot' failed >>>>>> make: *** [u-boot] Error 1 >>>>> >>>>> So why did it compile for me? Probably because the linker knows it doesn't >>>>> need 'clk_get_rate' since this branch will never be executed? >>>> Maybe you can try compile from clean build. Run "make mrproper" before compile. >>> >>> Of course I did that, and I just did it again. It *does* compile. >>> >>> Can anyone tell me a config/setup where it doesn't compile? Or does >>> this complain only come from reading the sources? >>> >> >> I see above error with k2g_evm_defconfig and compiler is: > > Ok, just tested that config and it works for me :-( After having a successful travis run for this on top of u-boot-socfpga/master, could you take this and re-send the PR of last week? Travis run is here: https://travis-ci.org/goldsimon/u-boot/builds/614187977 Regards, Simon > >> >> arm-linux-gnueabihf-gcc (GNU Toolchain for the A-profile Architecture >> 8.3-2019.03 (arm-rel-8.36)) 8.3.0 > > I'm using 6.3.0 from debian stretch, but have also tested my patch > with newest Ubuntu (which has a 9.x cross compiler). > > So while I think that difference is disturbing, Maybe it's really best > to inline-define all clock functions as you mentioned to Patrick > in the other thread... > > Regards, > Simon > >> >> Regards >> Vignesh >> >> >>> Regards, >>> Simon >>> >>>> >>>> Regards >>>> Ley Foon >>>>> >>>>> Regards, >>>>> Simon >>>>> >>>>>> >>>>>> Regards >>>>>> Vignesh >>>>>> >>>>>>> That should be enough, no? And yes, I did test this on the current >>>>>>> state of >>>>>>> gen5 which does not have a CLK driver, yet. >>>>>>> >>>>>>> Regards, >>>>>>> Simon >>>>>>> >>>>>>>> >>>>>>>> Regards >>>>>>>> Ley Foon >>>>>>>> >>>>>>>>> + ret = clk_get_by_index(bus, 0, &clk); >>>>>>>>> + if (ret) { >>>>>>>>> +#ifdef CONFIG_CQSPI_REF_CLK >>>>>>>>> + plat->ref_clk_hz = CONFIG_CQSPI_REF_CLK; #else >>>>>>>>> + return ret; >>>>>>>>> +#endif >>>>>>>>> + } else { >>>>>>>>> + plat->ref_clk_hz = clk_get_rate(&clk); >>>>>>>>> + clk_free(&clk); >>>>>>>>> + if (IS_ERR_VALUE(plat->ref_clk_hz)) >>>>>>>>> + return plat->ref_clk_hz; >>>>>>>>> + } >>>>>>>>> + >>>>>>>>> debug("%s: regbase=%p ahbbase=%p max-frequency=%d page- >>>>>>>>> size=%d\n", >>>>>>>>> __func__, plat->regbase, plat->ahbbase, plat->max_hz, >>>>>>>>> plat->page_size); >>>>>>>>> diff --git a/drivers/spi/cadence_qspi.h >>>>>>>>> b/drivers/spi/cadence_qspi.h index 20cceca239..99dee75bbd 100644 >>>>>>>>> --- a/drivers/spi/cadence_qspi.h >>>>>>>>> +++ b/drivers/spi/cadence_qspi.h >>>>>>>>> @@ -16,6 +16,7 @@ >>>>>>>>> #define CQSPI_READ_CAPTURE_MAX_DELAY 16 >>>>>>>>> >>>>>>>>> struct cadence_spi_platdata { >>>>>>>>> + unsigned int ref_clk_hz; >>>>>>>>> unsigned int max_hz; >>>>>>>>> void *regbase; >>>>>>>>> void *ahbbase; >>>>>>>>> -- >>>>>>>>> 2.20.1 >>>>>>>> >> >> -- >> Regards >> Vignesh