| Message ID | 20240206114349.32197-1-pmalgujar@marvell.com |
|---|---|
| Headers | show |
| Series | i2c: thunderx: Marvell thunderx i2c changes | expand |
Hi Piyush, On Tue, Feb 06, 2024 at 03:43:46AM -0800, Piyush Malgujar wrote: > From: Suneel Garapati <sgarapati@marvell.com> > > Handle changes to clock divisor logic for OcteonTX2 SoC family using > subsystem ID and using default reference clock source as 100MHz. > > Signed-off-by: Suneel Garapati <sgarapati@marvell.com> > Signed-off-by: Piyush Malgujar <pmalgujar@marvell.com> > Acked-by: Andi Shyti <andi.shyti@kernel.org> > --- > MAINTAINERS | 1 + > drivers/i2c/busses/i2c-octeon-core.c | 29 ++++++++++++++++++++---- > drivers/i2c/busses/i2c-octeon-core.h | 17 ++++++++++++++ > drivers/i2c/busses/i2c-thunderx-pcidrv.c | 7 ++++++ > 4 files changed, 50 insertions(+), 4 deletions(-) > > diff --git a/MAINTAINERS b/MAINTAINERS > index 960512bec42885c0f1632a7c90851c3d32fbf20e..92b0a55c36e41cf54c7cbf52576d5424b591aa31 100644 > --- a/MAINTAINERS > +++ b/MAINTAINERS > @@ -4725,6 +4725,7 @@ F: drivers/net/wireless/ath/carl9170/ > > CAVIUM I2C DRIVER > M: Robert Richter <rric@kernel.org> > +M: Suneel Garapati <sgarapati@marvell.com> Suneel doesn't have much of a history here, though. Can I have an ack from Robert? In any case, can you please put this change outside from the series as it would take a different path. Thanks, Andi
Hi Suneel and Piyush, ... > void octeon_i2c_set_clock(struct octeon_i2c *i2c) > { > int tclk, thp_base, inc, thp_idx, mdiv_idx, ndiv_idx, foscl, diff; > - int thp = 0x18, mdiv = 2, ndiv = 0, delta_hz = 1000000; > + int mdiv_min = 2; why is this int while all the other variables moved to be "unsigned int"? > + /* starting value on search for lowest diff */ > + const int huge_delta = INITIAL_DELTA_HZ; this is not needed. > + /* > + * Find divisors to produce target frequency, start with large delta > + * to cover wider range of divisors, note thp = TCLK half period. > + */ > + unsigned int thp = 0x18, mdiv = 2, ndiv = 0, delta_hz = huge_delta; > + > + if (octeon_i2c_is_otx2(to_pci_dev(i2c->dev))) { you could save this in a boolean at the beginning and keep using it. > + thp = 0x3; as you are here, can we give a meaningful define to 0x18 and 0x3? > + mdiv_min = 0; > + } > > for (ndiv_idx = 0; ndiv_idx < 8 && delta_hz != 0; ndiv_idx++) { > /* > * An mdiv value of less than 2 seems to not work well > * with ds1337 RTCs, so we constrain it to larger values. > */ > - for (mdiv_idx = 15; mdiv_idx >= 2 && delta_hz != 0; mdiv_idx--) { > + for (mdiv_idx = 15; mdiv_idx >= mdiv_min && delta_hz != 0; mdiv_idx--) { > /* > * For given ndiv and mdiv values check the > * two closest thp values. > */ > tclk = i2c->twsi_freq * (mdiv_idx + 1) * 10; > tclk *= (1 << ndiv_idx); > - thp_base = (i2c->sys_freq / (tclk * 2)) - 1; > + if (octeon_i2c_is_otx2(to_pci_dev(i2c->dev))) > + thp_base = (i2c->sys_freq / tclk) - 2; > + else > + thp_base = (i2c->sys_freq / (tclk * 2)) - 1; > > for (inc = 0; inc <= 1; inc++) { > thp_idx = thp_base + inc; > if (thp_idx < 5 || thp_idx > 0xff) > continue; > > - foscl = i2c->sys_freq / (2 * (thp_idx + 1)); > + if (octeon_i2c_is_otx2(to_pci_dev(i2c->dev))) > + foscl = i2c->sys_freq / (thp_idx + 2); > + else > + foscl = i2c->sys_freq / > + (2 * (thp_idx + 1)); I need to trust you on this. > foscl = foscl / (1 << ndiv_idx); > foscl = foscl / (mdiv_idx + 1) / 10; > diff = abs(foscl - i2c->twsi_freq); > + /* Use it if smaller diff from target */ can you please expand? > if (diff < delta_hz) { > delta_hz = diff; > thp = thp_idx; > diff --git a/drivers/i2c/busses/i2c-octeon-core.h b/drivers/i2c/busses/i2c-octeon-core.h > index 9bb9f64fdda0392364638ecbaafe3fab5612baf6..694c24cecb7b144c1021549d1661b040c21bb998 100644 > --- a/drivers/i2c/busses/i2c-octeon-core.h > +++ b/drivers/i2c/busses/i2c-octeon-core.h > @@ -7,6 +7,7 @@ > #include <linux/i2c-smbus.h> > #include <linux/io.h> > #include <linux/kernel.h> > +#include <linux/pci.h> > > /* Controller command patterns */ > #define SW_TWSI_V BIT_ULL(63) /* Valid bit */ > @@ -98,6 +99,8 @@ struct octeon_i2c_reg_offset { > #define TWSI_INT(x) (x->roff.twsi_int) > #define SW_TWSI_EXT(x) (x->roff.sw_twsi_ext) > > +#define INITIAL_DELTA_HZ 1000000 This define doesn't need to be in the header, right? Can we move it in the c file? Andi
Hi, On Tue, Feb 06, 2024 at 03:43:48AM -0800, Piyush Malgujar wrote: > From: Suneel Garapati <sgarapati@marvell.com> > > Status code 0xF0 refers to expiry of TWSI controller > access watchdog and needs bus monitor reset using MODE > register. Please describe what you did, why you did it and how you did it. Thanks, Andi