Skip to contentSkip to conversation

memory: tegra: Remove clients SID override programming

PATCH Superseded Ashish Mhetre 2 participants 5 messages Nov 22, 2022 – Nov 25, 2022

Subsystems (2)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 memory: tegra: Remove clients SID override programming

On newer Tegra releases, early boot SID override programming and SID override programming during resume is handled by bootloader. Also, SID override is programmed on-demand during probe_finalize() call of IOMMU which is done in tegra186_mc_client_sid_override() in this same file. This function does it more correctly by checking if write is permitted on SID override register. It also checks if SID override register is already written with correct value and skips re-writing it in that case.
Hence, removing the SID override programming of all clients.

Fixes 393d66fd2cac ("memory: tegra: Implement SID override programming")

---
Changes in v2:
- After offline discussions with Thierry and Jonathan, removed SID override programming during resume as well.

 drivers/memory/tegra/tegra186.c | 36 ---------------------------------
 1 file changed, 36 deletions(-)
drivers/memory/tegra/tegra186.c +0 -36
index 62477e592bf5..7bb73f06fad3 100644
@@ -22,32 +22,6 @@
 #define MC_SID_STREAMID_SECURITY_WRITE_ACCESS_DISABLED BIT(16)
 #define MC_SID_STREAMID_SECURITY_OVERRIDE BIT(8)
 
-static void tegra186_mc_program_sid(struct tegra_mc *mc)
-{
-	unsigned int i;
-
-	for (i = 0; i < mc->soc->num_clients; i++) {
-		const struct tegra_mc_client *client = &mc->soc->clients[i];
-		u32 override, security;
-
-		override = readl(mc->regs + client->regs.sid.override);
-		security = readl(mc->regs + client->regs.sid.security);
-
-		dev_dbg(mc->dev, "client %s: override: %x security: %x\n",
-			client->name, override, security);
-
-		dev_dbg(mc->dev, "setting SID %u for %s\n", client->sid,
-			client->name);
-		writel(client->sid, mc->regs + client->regs.sid.override);
-
-		override = readl(mc->regs + client->regs.sid.override);
-		security = readl(mc->regs + client->regs.sid.security);
-
-		dev_dbg(mc->dev, "client %s: override: %x security: %x\n",
-			client->name, override, security);
-	}
-}
-
 static int tegra186_mc_probe(struct tegra_mc *mc)
 {
 	struct platform_device *pdev = to_platform_device(mc->dev);
@@ -85,8 +59,6 @@ static int tegra186_mc_probe(struct tegra_mc *mc)
 	if (err < 0)
 		return err;
 
-	tegra186_mc_program_sid(mc);
-
 	return 0;
 }
 
@@ -95,13 +67,6 @@ static void tegra186_mc_remove(struct tegra_mc *mc)
 	of_platform_depopulate(mc->dev);
 }
 
-static int tegra186_mc_resume(struct tegra_mc *mc)
-{
-	tegra186_mc_program_sid(mc);
-
-	return 0;
-}
-
 #if IS_ENABLED(CONFIG_IOMMU_API)
 static void tegra186_mc_client_sid_override(struct tegra_mc *mc,
 					    const struct tegra_mc_client *client,
@@ -173,7 +138,6 @@ static int tegra186_mc_probe_device(struct tegra_mc *mc, struct device *dev)
 const struct tegra_mc_ops tegra186_mc_ops = {
 	.probe = tegra186_mc_probe,
 	.remove = tegra186_mc_remove,
-	.resume = tegra186_mc_resume,
 	.probe_device = tegra186_mc_probe_device,
 	.handle_irq = tegra30_mc_handle_irq,
 };
Signature · 1 line
2.17.1
#1 lore raw
Quoted 11 lines · Ashish Mhetre

On newer Tegra releases, early boot SID override programming and SID override programming during resume is handled by bootloader. Also, SID override is programmed on-demand during probe_finalize() call of IOMMU which is done in tegra186_mc_client_sid_override() in this same file. This function does it more correctly by checking if write is permitted on SID override register. It also checks if SID override register is already written with correct value and skips re-writing it in that case.
Hence, removing the SID override programming of all clients.

Fixes 393d66fd2cac ("memory: tegra: Implement SID override programming")

I could not get from commit msg what is the bug being fixed. You just said "more correctly", but usually things are either correct or not. What are visible effects of the bug?

Otherwise it sounds more like optimization or a bit better approach, but not a bugfix.

Best regards,
Krzysztof

#2 lore raw
Quoted 19 lines · Krzysztof Kozlowski

External email: Use caution opening links or attachments

Quoted 11 lines · Ashish Mhetre

On newer Tegra releases, early boot SID override programming and SID override programming during resume is handled by bootloader. Also, SID override is programmed on-demand during probe_finalize() call of IOMMU which is done in tegra186_mc_client_sid_override() in this same file. This function does it more correctly by checking if write is permitted on SID override register. It also checks if SID override register is already written with correct value and skips re-writing it in that case.
Hence, removing the SID override programming of all clients.

Fixes 393d66fd2cac ("memory: tegra: Implement SID override programming")

I could not get from commit msg what is the bug being fixed. You just said "more correctly", but usually things are either correct or not. What are visible effects of the bug?

Quoted 5 lines

Otherwise it sounds more like optimization or a bit better approach, but not a bugfix.

Best regards,
Krzysztof

Thanks for the review. In the function tegra186_mc_program_sid() which is getting removed, SID override register of all clients is written without checking if secure firmware has allowed write on it or not. If write is disabled by secure firmware then it can lead to errors coming from secure firmware and hang in kernel boot. So, that's a possible bug. Also, it's an optimization over current approach because it saves time by removing re-writing of these SID override registers as in new Tegra releases SID override of all clients is programmed by bootloader. So, MC driver don't need to program them again.

Thanks,
Ashish Mhetre

#3 lore raw
Quoted 24 lines · Ashish Mhetre
Quoted 22 lines · Krzysztof Kozlowski

External email: Use caution opening links or attachments

Quoted 11 lines · Ashish Mhetre

On newer Tegra releases, early boot SID override programming and SID override programming during resume is handled by bootloader. Also, SID override is programmed on-demand during probe_finalize() call of IOMMU which is done in tegra186_mc_client_sid_override() in this same file. This function does it more correctly by checking if write is permitted on SID override register. It also checks if SID override register is already written with correct value and skips re-writing it in that case.
Hence, removing the SID override programming of all clients.

Fixes 393d66fd2cac ("memory: tegra: Implement SID override programming")

I could not get from commit msg what is the bug being fixed. You just said "more correctly", but usually things are either correct or not. What are visible effects of the bug?

Otherwise it sounds more like optimization or a bit better approach, but not a bugfix.

Quoted 8 lines
Quoted 2 lines

Best regards,
Krzysztof

Thanks for the review. In the function tegra186_mc_program_sid() which is getting removed, SID override register of all clients is written without checking if secure firmware has allowed write on it or not. If write is disabled by secure firmware then it can lead to errors coming from secure firmware and hang in kernel boot. So, that's a possible bug.

Please add it to commit msg, because it justifies Fixes tag and probably backport.

Quoted 4 lines

Also, it's an optimization over current approach because it saves time by removing re-writing of these SID override registers as in new Tegra releases SID override of all clients is programmed by bootloader. So, MC driver don't need to program them again.

Best regards,
Krzysztof

#4 lore raw
Quoted 29 lines · Krzysztof Kozlowski

External email: Use caution opening links or attachments

Quoted 25 lines · Ashish Mhetre
Quoted 24 lines · Krzysztof Kozlowski

External email: Use caution opening links or attachments

Quoted 11 lines · Ashish Mhetre

On newer Tegra releases, early boot SID override programming and SID override programming during resume is handled by bootloader. Also, SID override is programmed on-demand during probe_finalize() call of IOMMU which is done in tegra186_mc_client_sid_override() in this same file. This function does it more correctly by checking if write is permitted on SID override register. It also checks if SID override register is already written with correct value and skips re-writing it in that case.
Hence, removing the SID override programming of all clients.

Fixes 393d66fd2cac ("memory: tegra: Implement SID override programming")

I could not get from commit msg what is the bug being fixed. You just said "more correctly", but usually things are either correct or not. What are visible effects of the bug?

Otherwise it sounds more like optimization or a bit better approach, but not a bugfix.

Best regards,
Krzysztof

Quoted 7 lines
Quoted 5 lines

Thanks for the review. In the function tegra186_mc_program_sid() which is getting removed, SID override register of all clients is written without checking if secure firmware has allowed write on it or not. If write is disabled by secure firmware then it can lead to errors coming from secure firmware and hang in kernel boot. So, that's a possible bug.

Please add it to commit msg, because it justifies Fixes tag and probably backport.

Sure, I will send V3 with updated commit message. Thank you.

Quoted 7 lines
Quoted 4 lines

Also, it's an optimization over current approach because it saves time by removing re-writing of these SID override registers as in new Tegra releases SID override of all clients is programmed by bootloader. So, MC driver don't need to program them again.

Best regards,
Krzysztof