Message ID | 20201008232154.94221-1-hskinnemoen@google.com |
---|---|
Headers | show |
Series | Additional NPCM7xx features, devices and tests | expand |
On Fri, 9 Oct 2020 at 00:22, Havard Skinnemoen <hskinnemoen@google.com> wrote: > > From: Hao Wu <wuhaotsh@google.com> > > The watchdog is part of NPCM7XX's timer module. Its behavior is > controlled by the WTCR register in the timer. > > When enabled, the watchdog issues an interrupt signal after a pre-set > amount of cycles, and issues a reset signal shortly after that. > > Reviewed-by: Tyrone Ting <kfting@nuvoton.com> > Signed-off-by: Hao Wu <wuhaotsh@google.com> > Signed-off-by: Havard Skinnemoen <hskinnemoen@google.com> > --- > include/hw/misc/npcm7xx_clk.h | 9 + > include/hw/timer/npcm7xx_timer.h | 43 ++- > hw/arm/npcm7xx.c | 11 + > hw/misc/npcm7xx_clk.c | 20 ++ > hw/timer/npcm7xx_timer.c | 275 +++++++++++++++---- > tests/qtest/npcm7xx_watchdog_timer-test.c | 313 ++++++++++++++++++++++ > MAINTAINERS | 1 + > tests/qtest/meson.build | 1 + > 8 files changed, 620 insertions(+), 53 deletions(-) > create mode 100644 tests/qtest/npcm7xx_watchdog_timer-test.c > > diff --git a/include/hw/misc/npcm7xx_clk.h b/include/hw/misc/npcm7xx_clk.h > index cdcc9e8534..483028cf63 100644 > --- a/include/hw/misc/npcm7xx_clk.h > +++ b/include/hw/misc/npcm7xx_clk.h > @@ -42,6 +42,15 @@ typedef struct NPCM7xxCLKState { > int64_t ref_ns; > } NPCM7xxCLKState; > > +/** > + * npcm7xx_clk_perform_watchdog_reset - Perform watchdog reset action generated > + * by a watchdog event. > + * @clk: The clock module that connects to the watchdog. > + * @watchdog_index: The index of the watchdog that triggered the reset action. > + */ > +void npcm7xx_clk_perform_watchdog_reset(NPCM7xxCLKState *clk, > + int watchdog_index); It looks like you're using this as a mechanism for having the watchdog module signal to the clock module that it needs to do a reset. The usual way I'd implement a cross-device link like that would be to use a qemu_irq for it -- the receiving end creates a named GPIO input, the sending end has a named GPIO output, and the SoC that creates both devices also wires the GPIO link from one to the other. (You can send an arbitrary integer down a qemu_irq, not just an on/off, so you could do this either with an array of N GPIOs, one per watchdog, or with just one that you send an index value down.) I have a handful of nits below but other than the above issue this patch looks good. > +static void npcm7xx_watchdog_timer_reset_cycles(NPCM7xxWatchdogTimer *t, > + int64_t cycles) > +{ > + uint32_t prescaler = npcm7xx_watchdog_timer_prescaler(t); > + int64_t ns = (NANOSECONDS_PER_SECOND / NPCM7XX_TIMER_REF_HZ) * cycles; > + > + /* > + * The reset function always clear the current timer. The caller to the > + * this needs to decide whether to start the watchdog timer based on "always clears". "The caller of this". > + * specific flag in WTCR. > + */ > + npcm7xx_timer_clear(&t->base_timer); > + > + ns *= prescaler; > + t->base_timer.remaining_ns = ns; > +} > @@ -259,9 +333,47 @@ static void npcm7xx_timer_write_tisr(NPCM7xxTimerCtrlState *s, uint32_t value) > if (value & (1U << i)) { > npcm7xx_timer_check_interrupt(&s->timer[i]); > } > + Stray extra blank line? > } > } > > + /* > + * Set WTCLK to 1(default) and resets all flags except WTRF. > + * WTRF is not reset during a core domain reset. > + */ "and reset all flags" (or "Sets", if you prefer) + s->watchdog_timer.wtcr = 0x00000400 | (s->watchdog_timer.wtcr & > + NPCM7XX_WTCR_WTRF); > +} > + > +static void npcm7xx_watchdog_timer_expired(void *opaque) > +{ > + NPCM7xxWatchdogTimer *t = opaque; > + > + if (t->wtcr & NPCM7XX_WTCR_WTE) { > + if (t->wtcr & NPCM7XX_WTCR_WTIF) { > + if (t->wtcr & NPCM7XX_WTCR_WTRE) { > + t->wtcr |= NPCM7XX_WTCR_WTRF; > + /* send reset signal to CLK module*/ > + npcm7xx_clk_perform_watchdog_reset(t->ctrl->clk, > + t->ctrl->index); > + } > + } else { > + t->wtcr |= NPCM7XX_WTCR_WTIF; > + if (t->wtcr & NPCM7XX_WTCR_WTIE) > + /* send interrupt */ > + qemu_irq_raise(t->irq); This if() is missing its mandatory braces. (Not sure why checkpatch doesn't catch this.) > + npcm7xx_watchdog_timer_reset_cycles(t, > + NPCM7XX_WATCHDOG_INTERRUPT_TO_RESET_CYCLES); > + npcm7xx_timer_start(&t->base_timer); > + } > + } > } thanks -- PMM
On Fri, 9 Oct 2020 at 00:22, Havard Skinnemoen <hskinnemoen@google.com> wrote: > > This is an update to the initial NPCM7xx patch series adding > > - A timer test that found several issues that were fixed in the final version > of the series (see > https://www.mail-archive.com/qemu-devel@nongnu.org/msg739516.html). > - Watchdog timer support. This makes the reboot command work. > - Random Number Generator device. > - USB Host Controllers. > - GPIO Controllers. > > The watchdog was implemented by my new teammate Hao Wu. Expect to see more > patches from him in the near future. > > This series has also been pushed to the npcm7xx-5.2-update branch of my github > repository at > > https://github.com/hskinnemoen/qemu > > Again, thanks a lot for reviewing! I'm going to take patch 1 (the timer test case) into my target-arm.next queue; I've sent some review comments on the other parts of the series. thanks -- PMM