On 9/22/25 5:33 PM, Geert Uytterhoeven wrote:
Hello Geert,
On Mon, 22 Sept 2025 at 17:17, Marek Vasut <marek.vasut@xxxxxxxxxxx> wrote:
On 9/22/25 12:10 PM, Geert Uytterhoeven wrote:
I have instead posted what I think are proper fixes for that SError:
PCI: rcar-gen4: Add missing 1ms delay after PWR reset assertion
https://patchwork.kernel.org/project/linux-pci/patch/20250918030058.330960-1-marek.vasut+renesas@xxxxxxxxxxx/
I used v3 instead.
While that patch seems to fix the SError after a hard reset (hardware
reset), it is not sufficient after a soft reset (typing "reboot").
clk: renesas: cpg-mssr: Add missing 1ms delay into reset toggle callback
https://patchwork.kernel.org/project/linux-clk/patch/20250918030552.331389-1-marek.vasut+renesas@xxxxxxxxxxx/
This does not fix the SError, as expected (pcie-rcar-gen4.c does not
call reset_control_reset(), but reset_control_{,de}assert()).
clk: renesas: cpg-mssr: Read back reset registers to assure values latched
https://patchwork.kernel.org/project/linux-clk/patch/20250918030723.331634-1-marek.vasut+renesas@xxxxxxxxxxx/
I used v2 instead, which seems to fix the SError.
Those three patches have to be used together, and this inverted break
condition fix should be applied too.
The first two are corrections which align the code behavior with
reference manual. This inverted break fix is another correction. The
last patch in the list above actually fixes the asynchronized reset
behavior and turns it into synchronized reset behavior, therefore fixing
the SError in the process.
FTR, I always had the inverted break condition fix applied.
All 3 fixes on top should be fine.
Maybe I can finally properly deserve your TB on this patch with this
option (C) , all three patches applied.
