Re: [PATCH 01/10] ARM: ti81xx: Prevent USB being put into idle

From: Andrew Goodbody

Date: Mon Oct 19 2026 - 05:34:36 EST


On 17/10/2026 07:53, Andreas Kemnade wrote:
On Fri, 16 Oct 2026 16:41:16 +0200
Ettore Chimenti <ettore.chimenti@xxxxxxxxxx> wrote:

From: Andrew Goodbody <andrew.goodbody@xxxxxxxxxx>

There is an issue with USB being put into idle with DMA enabled and that
is mostly prevented with usbcore.autosuspend=-1 on the kernel command
line. Stop the ti,hwmod code from putting it into idle by setting the
appropriate flag. This stops the log entry 'clk:0058:0: failed to
disable' being generated.

So the module cannot be put into idle in certain conditions. Why do you
disable idle under all circumstances and not e.g. adding some
runtime_resume/suspend pair somewhere?

Signed-off-by: Andrew Goodbody <andrew.goodbody@xxxxxxxxxx>
Signed-off-by: Ettore Chimenti <ettore.chimenti@xxxxxxxxxx>
---
arch/arm/mach-omap2/omap_hwmod_81xx_data.c | 1 +
1 file changed, 1 insertion(+)

diff --git a/arch/arm/mach-omap2/omap_hwmod_81xx_data.c b/arch/arm/mach-omap2/omap_hwmod_81xx_data.c
index 9b5c728fb7da..2f55dc32e150 100644
--- a/arch/arm/mach-omap2/omap_hwmod_81xx_data.c
+++ b/arch/arm/mach-omap2/omap_hwmod_81xx_data.c
@@ -650,6 +650,7 @@ static struct omap_hwmod dm816x_usbss_hwmod = {
.name = "usb_otg_hs",
.clkdm_name = "default_l3_slow_clkdm",
.main_clk = "sysclk6_ck",
+ .flags = HWMOD_NO_IDLE,

and why do we need this here and not in the ti-sysc node?

Regards,
Andreas

Sorry for the noise, this can be dropped. It is not needed. It was a workaround for another problem that has since been fixed.

Andrew