Skip to contentSkip to conversation

memory: tegra: Check whether reset is already asserted

PATCH In review Dmitry Osipenko 2 participants 5 messages Jan 19, 2021 – Jan 23, 2021

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 memory: tegra: Check whether reset is already asserted

Check whether memory client reset is already asserted in order to prevent DMA-flush error on trying to re-assert an already asserted reset.

This becomes a problem once PMC GENPD is enabled to use memory resets since GENPD will get a error and fail to toggle power domain. PMC GENPDs can't be toggled safely without holding memory reset on Tegra and we're about to fix this.

---
 drivers/memory/tegra/mc.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/memory/tegra/mc.c +7 -0
index 44064de962c2..a21163ccadc4 100644
@@ -176,6 +176,13 @@ static int tegra_mc_hotreset_assert(struct reset_controller_dev *rcdev,
 	if (!rst_ops)
 		return -ENODEV;
 
+	/* DMA flushing will fail if reset is already asserted */
+	if (rst_ops->reset_status) {
+		/* check whether reset is asserted */
+		if (rst_ops->reset_status(mc, rst))
+			return 0;
+	}
+
 	if (rst_ops->block_dma) {
 		/* block clients DMA requests */
 		err = rst_ops->block_dma(mc, rst);
Signature · 1 line
2.29.2
#1 lore raw
Quoted 10 lines · Dmitry Osipenko

Check whether memory client reset is already asserted in order to prevent DMA-flush error on trying to re-assert an already asserted reset.

This becomes a problem once PMC GENPD is enabled to use memory resets since GENPD will get a error and fail to toggle power domain. PMC GENPDs can't be toggled safely without holding memory reset on Tegra and we're about to fix this.

Don't add your own Tested-by. Author has to test it. However this is a v1 and already three other people added tests. Where are the emails?

Quoted 7 lines
---
 drivers/memory/tegra/mc.c | 7 +++++++
 1 file changed, 7 insertions(+)

Best regards,
Krzysztof

#2 lore raw

23.01.2021 00:40, Krzysztof Kozlowski пишет:

Quoted 9 lines · Dmitry Osipenko
Quoted 8 lines

Check whether memory client reset is already asserted in order to prevent DMA-flush error on trying to re-assert an already asserted reset.

This becomes a problem once PMC GENPD is enabled to use memory resets since GENPD will get a error and fail to toggle power domain. PMC GENPDs can't be toggled safely without holding memory reset on Tegra and we're about to fix this.

Quoted 5 lines
Quoted 2 lines

Don't add your own Tested-by. Author has to test it. However this is a v1 and already three other people added tests. Where are the emails?

The test results were given on #tegra irc.

Author doesn't have to test, from time to time I'm helping people who can't make a patch with fixing obvious bugs for the h/w that I don't have access to.

Anyways, I'll make v2 with myself removed if you prefer that, thanks.

#3 lore raw
Quoted 13 lines · Dmitry Osipenko

23.01.2021 00:40, Krzysztof Kozlowski пишет:

Quoted 12 lines · Dmitry Osipenko
Quoted 10 lines

Check whether memory client reset is already asserted in order to prevent DMA-flush error on trying to re-assert an already asserted reset.

This becomes a problem once PMC GENPD is enabled to use memory resets since GENPD will get a error and fail to toggle power domain. PMC GENPDs can't be toggled safely without holding memory reset on Tegra and we're about to fix this.

Quoted 8 lines
Quoted 2 lines

Don't add your own Tested-by. Author has to test it. However this is a v1 and already three other people added tests. Where are the emails?

The test results were given on #tegra irc.

Author doesn't have to test, from time to time I'm helping people who can't make a patch with fixing obvious bugs for the h/w that I don't have access to.

By default, author's patches are assumed to be tested, otherwise we would be putting own Tested-by in most of the cases (ideally: in all). Just because you do not see author's Tested-by everywhere it does not mean that author did not test it...

Testing is also a requirement:
https://elixir.bootlin.com/linux/latest/source/Documentation/process/5.Posting.rst#L35

Therefore if your non-trivial patches are not tested, please also mention it with RFT or after '---'.

Quoted 2 lines

Anyways, I'll make v2 with myself removed if you prefer that, thanks.

I applied it without your tested-by. If the tested platform is anyhow relevant, just mention it in the commit msg.

Best regards,
Krzysztof

#4 lore raw

23.01.2021 02:19, Krzysztof Kozlowski пишет:

Quoted 31 lines · Dmitry Osipenko
Quoted 13 lines

23.01.2021 00:40, Krzysztof Kozlowski пишет:

Quoted 12 lines · Dmitry Osipenko
Quoted 10 lines

Check whether memory client reset is already asserted in order to prevent DMA-flush error on trying to re-assert an already asserted reset.

This becomes a problem once PMC GENPD is enabled to use memory resets since GENPD will get a error and fail to toggle power domain. PMC GENPDs can't be toggled safely without holding memory reset on Tegra and we're about to fix this.

Quoted 8 lines
Quoted 2 lines

Don't add your own Tested-by. Author has to test it. However this is a v1 and already three other people added tests. Where are the emails?

The test results were given on #tegra irc.

Author doesn't have to test, from time to time I'm helping people who can't make a patch with fixing obvious bugs for the h/w that I don't have access to.

By default, author's patches are assumed to be tested, otherwise we would be putting own Tested-by in most of the cases (ideally: in all). Just because you do not see author's Tested-by everywhere it does not mean that author did not test it...

Testing is also a requirement:
https://elixir.bootlin.com/linux/latest/source/Documentation/process/5.Posting.rst#L35

Quoted 8 lines

Therefore if your non-trivial patches are not tested, please also mention it with RFT or after '---'.

Quoted 2 lines

Anyways, I'll make v2 with myself removed if you prefer that, thanks.

I applied it without your tested-by. If the tested platform is anyhow relevant, just mention it in the commit msg.

Alright, thank you. In this particular case the platform should be irrelevant.