| Message ID | 20190208123635.26755-1-masneyb@onstation.org |
|---|---|
| State | New |
| Headers | show |
| Series | [-next] spmi: pmic-arb: revert "validate type when mapping IRQ" | expand |
Hi Brian, I simply applied this patch to the pin control tree so we get it in with the rest of the patches. On Fri, Feb 8, 2019 at 1:36 PM Brian Masney <masneyb@onstation.org> wrote: > Validation of the IRQ type was added to spmi pmic-arb, however spmi-mpp > in device tree still uses IRQ_TYPE_NONE. This commit caused the > spmi-mpp probe to fail since platform_irq_count() would return 0. > Correct this by backing out the previous patch. > > Signed-off-by: Brian Masney <masneyb@onstation.org> > Fixes: 135ef21ab064 ("spmi: pmic-arb: validate type when mapping IRQ") > --- > Linus: I think that the better longer-term approach is to just embed the > IRQ counts similar to what is now in spmi-gpio. That is most certainly better. > Since we are getting > close to the end of the dev cycle, I think it is safer to just drop > the old patch for the time being. I'll commit to doing that work for the > next dev cycle, although I can send you that change this weekend if you > prefer that we go that route instead. It's fine like this. > My ssbi-gpio series does not have this issue. OK! Yours, Linus Walleij
diff --git a/drivers/spmi/spmi-pmic-arb.c b/drivers/spmi/spmi-pmic-arb.c index 316e2708f3ba..928759242e42 100644 --- a/drivers/spmi/spmi-pmic-arb.c +++ b/drivers/spmi/spmi-pmic-arb.c @@ -740,9 +740,9 @@ static int qpnpint_irq_domain_translate(struct irq_domain *d, } -static int qpnpint_irq_domain_map(struct spmi_pmic_arb *pmic_arb, - struct irq_domain *domain, unsigned int virq, - irq_hw_number_t hwirq, unsigned int type) +static void qpnpint_irq_domain_map(struct spmi_pmic_arb *pmic_arb, + struct irq_domain *domain, unsigned int virq, + irq_hw_number_t hwirq, unsigned int type) { irq_flow_handler_t handler; @@ -751,15 +751,11 @@ static int qpnpint_irq_domain_map(struct spmi_pmic_arb *pmic_arb, if (type & IRQ_TYPE_EDGE_BOTH) handler = handle_edge_irq; - else if (type & (IRQ_TYPE_LEVEL_LOW | IRQ_TYPE_LEVEL_HIGH)) - handler = handle_level_irq; else - return -EINVAL; + handler = handle_level_irq; irq_domain_set_info(domain, virq, hwirq, &pmic_arb_irqchip, pmic_arb, handler, NULL, NULL); - - return 0; } static int qpnpint_irq_domain_alloc(struct irq_domain *domain, @@ -776,12 +772,9 @@ static int qpnpint_irq_domain_alloc(struct irq_domain *domain, if (ret) return ret; - for (i = 0; i < nr_irqs; i++) { - ret = qpnpint_irq_domain_map(pmic_arb, domain, virq + i, - hwirq + i, type); - if (ret) - return ret; - } + for (i = 0; i < nr_irqs; i++) + qpnpint_irq_domain_map(pmic_arb, domain, virq + i, hwirq + i, + type); return 0; }
Validation of the IRQ type was added to spmi pmic-arb, however spmi-mpp in device tree still uses IRQ_TYPE_NONE. This commit caused the spmi-mpp probe to fail since platform_irq_count() would return 0. Correct this by backing out the previous patch. Signed-off-by: Brian Masney <masneyb@onstation.org> Fixes: 135ef21ab064 ("spmi: pmic-arb: validate type when mapping IRQ") --- Linus: I think that the better longer-term approach is to just embed the IRQ counts similar to what is now in spmi-gpio. Since we are getting close to the end of the dev cycle, I think it is safer to just drop the old patch for the time being. I'll commit to doing that work for the next dev cycle, although I can send you that change this weekend if you prefer that we go that route instead. My ssbi-gpio series does not have this issue. drivers/spmi/spmi-pmic-arb.c | 21 +++++++-------------- 1 file changed, 7 insertions(+), 14 deletions(-)