U-Boot CONFIG_TIMER migration deadline

Hey all,
I'm sending this email out to the main and custodian lists, and bcc'ing all of the listed maintainers for platforms that have a problem here. The migration deadline for switching from CONFIG_SYS_TIMER_RATE or CONFIG_SYS_TIMER_COUNTER and to using CONFIG_TIMER is the v2023.01 release. Now, per how I've behaved in the past, I'm not going to immediately remove these platforms. But it would be very good to jump on migrating them now, rather than waiting further. The full list (at least n so far as our Makefile check can determine) is: rpi rpi_0_w at91sam9g10ek_dataflash_cs0 at91sam9g10ek_dataflash_cs3 at91sam9g10ek_nandflash usb_a9263_dataflash ethernut5 meesc meesc_dataflash pm9261 pm9263 corvus smartweb axm taurus vexpress_ca9x4 vinco rpi_2 rpi_3_32b rpi_4_32b highbank alt blanche gose koelsch lager porter silk stout elgin-rv1108 evb-rv1108 socfpga_is1 socfpga_sr1500 socfpga_arria5 socfpga_cyclone5 socfpga_mcvevk socfpga_dbm_soc1 socfpga_socrates socfpga_secu1 socfpga_vining_fpga socfpga_de0_nano_soc socfpga_de1_soc socfpga_de10_nano socfpga_de10_standard socfpga_sockit dalmore cei-tk1-som jetson-tk1 nyan-big venice2 apalis-tk1 medcom-wide plutux tec paz00 trimslice harmony seaboard ventana colibri_t20 tec-ng beaver cardhu apalis_t30 colibri_t30 uniphier_ld4_sld8 uniphier_v7 cortina_presidio-asic-base cortina_presidio-asic-emmc cortina_presidio-asic-pnand r2dplus

Hey Tom,
I'm sending this email out to the main and custodian lists, and bcc'ing all of the listed maintainers for platforms that have a problem here. The migration deadline for switching from CONFIG_SYS_TIMER_RATE or CONFIG_SYS_TIMER_COUNTER and to using CONFIG_TIMER is the v2023.01 release. Now, per how I've behaved in the past, I'm not going to immediately remove these platforms. But it would be very good to jump on migrating them now, rather than waiting further. The full list (at least
Is there a good example commit that handles the migration?
n so far as our Makefile check can determine) is: rpi rpi_0_w at91sam9g10ek_dataflash_cs0 at91sam9g10ek_dataflash_cs3 at91sam9g10ek_nandflash usb_a9263_dataflash ethernut5 meesc meesc_dataflash pm9261 pm9263 corvus smartweb axm taurus vexpress_ca9x4 vinco rpi_2 rpi_3_32b rpi_4_32b highbank alt blanche gose koelsch lager porter silk stout elgin-rv1108 evb-rv1108 socfpga_is1 socfpga_sr1500 socfpga_arria5 socfpga_cyclone5 socfpga_mcvevk socfpga_dbm_soc1 socfpga_socrates socfpga_secu1 socfpga_vining_fpga socfpga_de0_nano_soc socfpga_de1_soc socfpga_de10_nano socfpga_de10_standard socfpga_sockit dalmore cei-tk1-som jetson-tk1 nyan-big venice2 apalis-tk1 medcom-wide plutux tec paz00 trimslice harmony seaboard ventana colibri_t20 tec-ng beaver cardhu apalis_t30 colibri_t30 uniphier_ld4_sld8 uniphier_v7 cortina_presidio-asic-base cortina_presidio-asic-emmc cortina_presidio-asic-pnand r2dplus
-- Tom

On Fri, Oct 14, 2022 at 02:32:41PM +0100, Peter Robinson wrote:
Hey Tom,
I'm sending this email out to the main and custodian lists, and bcc'ing all of the listed maintainers for platforms that have a problem here. The migration deadline for switching from CONFIG_SYS_TIMER_RATE or CONFIG_SYS_TIMER_COUNTER and to using CONFIG_TIMER is the v2023.01 release. Now, per how I've behaved in the past, I'm not going to immediately remove these platforms. But it would be very good to jump on migrating them now, rather than waiting further. The full list (at least
Is there a good example commit that handles the migration?
Good question! You'll need a driver under drivers/timer/ and for example in U-Boot drivers/timer/omap-timer.c matches (based on compatibles) with drivers/clocksource/timer-ti-dm-systimer.c and drivers/clocksource/timer-ti-dm.c in the Linux kernel. Otherwise, a git log drivers/timer/ shows various conversions of previous in-tree timers to this framework.

On 10/14/22 15:25, Tom Rini wrote:
Hey all,
I'm sending this email out to the main and custodian lists, and bcc'ing all of the listed maintainers for platforms that have a problem here. The migration deadline for switching from CONFIG_SYS_TIMER_RATE or CONFIG_SYS_TIMER_COUNTER and to using CONFIG_TIMER is the v2023.01 release. Now, per how I've behaved in the past, I'm not going to immediately remove these platforms. But it would be very good to jump on migrating them now, rather than waiting further. The full list (at least n so far as our Makefile check can determine) is: rpi rpi_0_w at91sam9g10ek_dataflash_cs0 at91sam9g10ek_dataflash_cs3 at91sam9g10ek_nandflash usb_a9263_dataflash ethernut5 meesc meesc_dataflash pm9261 pm9263 corvus smartweb axm taurus vexpress_ca9x4 vinco rpi_2 rpi_3_32b rpi_4_32b highbank alt blanche gose koelsch lager porter silk stout elgin-rv1108 evb-rv1108 socfpga_is1 socfpga_sr1500 socfpga_arria5 socfpga_cyclone5 socfpga_mcvevk socfpga_dbm_soc1 socfpga_socrates socfpga_secu1 socfpga_vining_fpga socfpga_de0_nano_soc socfpga_de1_soc socfpga_de10_nano socfpga_de10_standard socfpga_sockit dalmore cei-tk1-som jetson-tk1 nyan-big venice2 apalis-tk1 medcom-wide plutux tec paz00 trimslice harmony seaboard ventana colibri_t20 tec-ng beaver cardhu apalis_t30 colibri_t30 uniphier_ld4_sld8 uniphier_v7 cortina_presidio-asic-base cortina_presidio-asic-emmc cortina_presidio-asic-pnand r2dplus
Are you sure you didn't include machines which enable non-DM timer in SPL due to size limits?

On Fri, Oct 14, 2022 at 03:46:45PM +0200, Marek Vasut wrote:
On 10/14/22 15:25, Tom Rini wrote:
Hey all,
I'm sending this email out to the main and custodian lists, and bcc'ing all of the listed maintainers for platforms that have a problem here. The migration deadline for switching from CONFIG_SYS_TIMER_RATE or CONFIG_SYS_TIMER_COUNTER and to using CONFIG_TIMER is the v2023.01 release. Now, per how I've behaved in the past, I'm not going to immediately remove these platforms. But it would be very good to jump on migrating them now, rather than waiting further. The full list (at least n so far as our Makefile check can determine) is: rpi rpi_0_w at91sam9g10ek_dataflash_cs0 at91sam9g10ek_dataflash_cs3 at91sam9g10ek_nandflash usb_a9263_dataflash ethernut5 meesc meesc_dataflash pm9261 pm9263 corvus smartweb axm taurus vexpress_ca9x4 vinco rpi_2 rpi_3_32b rpi_4_32b highbank alt blanche gose koelsch lager porter silk stout elgin-rv1108 evb-rv1108 socfpga_is1 socfpga_sr1500 socfpga_arria5 socfpga_cyclone5 socfpga_mcvevk socfpga_dbm_soc1 socfpga_socrates socfpga_secu1 socfpga_vining_fpga socfpga_de0_nano_soc socfpga_de1_soc socfpga_de10_nano socfpga_de10_standard socfpga_sockit dalmore cei-tk1-som jetson-tk1 nyan-big venice2 apalis-tk1 medcom-wide plutux tec paz00 trimslice harmony seaboard ventana colibri_t20 tec-ng beaver cardhu apalis_t30 colibri_t30 uniphier_ld4_sld8 uniphier_v7 cortina_presidio-asic-base cortina_presidio-asic-emmc cortina_presidio-asic-pnand r2dplus
Are you sure you didn't include machines which enable non-DM timer in SPL due to size limits?
The list is based on our build warning logic, so no, I'm not 100% sure. If you know something above is using CONFIG_TIMER in full U-Boot but the old method in SPL, let me know and I'll see if I can rework the logic.
participants (3)
-
Marek Vasut
-
Peter Robinson
-
Tom Rini