Skip to contentSkip to conversation

net/mlx5: Convert PCI error values to generic errnos

PATCH In review Ilpo Järvinen 4 participants 8 messages Aug 14, 2023 – Aug 16, 2023

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 net/mlx5: Convert PCI error values to generic errnos

mlx5_pci_link_toggle() returns mix PCI specific error codes and generic errnos.

Convert the PCI specific error values to generic errno using pcibios_err_to_errno() before returning them.

Fixes eabe8e5e88f5 ("net/mlx5: Handle sync reset now event")
Fixes 212b4d7251c1 ("net/mlx5: Wait for firmware to enable CRS before pci_restore_state")

---

Maintainers beware, this will conflict with read+write -> set/clear_word fixes in pci.git/pcie-rmw. As such, it might be the easiest for Bjorn to take it instead of net people.

I wonder if these PCIBIOS_* error codes are useful at all? There's 1:1 mapping into errno values so no information loss if the functions would just return errnos directly. Perhaps this is just legacy nobody has bothered to remove? If nobody opposes, I could take a look at getting rid of them.

---
 drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c +6 -6
index 4804990b7f22..0afd9dbfc471 100644
@@ -371,7 +371,7 @@ static int mlx5_pci_link_toggle(struct mlx5_core_dev *dev)
 
 	err = pci_read_config_word(dev->pdev, PCI_DEVICE_ID, &dev_id);
 	if (err)
-		return err;
+		return pcibios_err_to_errno(err);
 	err = mlx5_check_dev_ids(dev, dev_id);
 	if (err)
 		return err;
@@ -386,16 +386,16 @@ static int mlx5_pci_link_toggle(struct mlx5_core_dev *dev)
 	/* PCI link toggle */
 	err = pci_read_config_word(bridge, cap + PCI_EXP_LNKCTL, &reg16);
 	if (err)
-		return err;
+		return pcibios_err_to_errno(err);
 	reg16 |= PCI_EXP_LNKCTL_LD;
 	err = pci_write_config_word(bridge, cap + PCI_EXP_LNKCTL, reg16);
 	if (err)
-		return err;
+		return pcibios_err_to_errno(err);
 	msleep(500);
 	reg16 &= ~PCI_EXP_LNKCTL_LD;
 	err = pci_write_config_word(bridge, cap + PCI_EXP_LNKCTL, reg16);
 	if (err)
-		return err;
+		return pcibios_err_to_errno(err);
 
 	/* Check link */
 	if (!bridge->link_active_reporting) {
@@ -408,7 +408,7 @@ static int mlx5_pci_link_toggle(struct mlx5_core_dev *dev)
 	do {
 		err = pci_read_config_word(bridge, cap + PCI_EXP_LNKSTA, &reg16);
 		if (err)
-			return err;
+			return pcibios_err_to_errno(err);
 		if (reg16 & PCI_EXP_LNKSTA_DLLLA)
 			break;
 		msleep(20);
@@ -426,7 +426,7 @@ static int mlx5_pci_link_toggle(struct mlx5_core_dev *dev)
 	do {
 		err = pci_read_config_word(dev->pdev, PCI_DEVICE_ID, &reg16);
 		if (err)
-			return err;
+			return pcibios_err_to_errno(err);
 		if (reg16 == dev_id)
 			break;
 		msleep(20);
Signature · 1 line
2.30.2
#1 lore raw
Quoted 10 lines · Ilpo Järvinen

mlx5_pci_link_toggle() returns mix PCI specific error codes and generic errnos.

Convert the PCI specific error values to generic errno using pcibios_err_to_errno() before returning them.

Fixes eabe8e5e88f5 ("net/mlx5: Handle sync reset now event")
Fixes 212b4d7251c1 ("net/mlx5: Wait for firmware to enable CRS before pci_restore_state")
Quoted 5 lines

---

Maintainers beware, this will conflict with read+write -> set/clear_word fixes in pci.git/pcie-rmw. As such, it might be the easiest for Bjorn to take it instead of net people.

I provisionally rebased and applied it on pci/pcie-rmw. Take a look and make sure I didn't botch it -- I also found a case in mlx5_check_dev_ids() that looks like it needs the same conversion.

The commit as applied is below.

If networking folks would prefer to take this, let me know and I can drop it.

Quoted 4 lines

I wonder if these PCIBIOS_* error codes are useful at all? There's 1:1 mapping into errno values so no information loss if the functions would just return errnos directly. Perhaps this is just legacy nobody has bothered to remove? If nobody opposes, I could take a look at getting rid of them.

I don't think the PCIBIOS error codes are very useful outside of arch/x86. They're returned by x86 PCIBIOS functions, and I think we still use those calls, but I don't think there's value in exposing the x86 error codes outside arch/x86. Looks like a big job to clean it up though ;)

Quoted 47 lines
---
 drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c +5 -5
index 4804990b7f22..0afd9dbfc471 100644
@@ -371,7 +371,7 @@ static int mlx5_pci_link_toggle(struct mlx5_core_dev *dev)
 
 	err = pci_read_config_word(dev->pdev, PCI_DEVICE_ID, &dev_id);
 	if (err)
-		return err;
+		return pcibios_err_to_errno(err);
 	err = mlx5_check_dev_ids(dev, dev_id);
 	if (err)
 		return err;
@@ -386,16 +386,16 @@ static int mlx5_pci_link_toggle(struct mlx5_core_dev *dev)
 	/* PCI link toggle */
 	err = pci_read_config_word(bridge, cap + PCI_EXP_LNKCTL, &reg16);
 	if (err)
-		return err;
+		return pcibios_err_to_errno(err);
 	reg16 |= PCI_EXP_LNKCTL_LD;
 	err = pci_write_config_word(bridge, cap + PCI_EXP_LNKCTL, reg16);
 	if (err)
-		return err;
+		return pcibios_err_to_errno(err);
 	msleep(500);
 	reg16 &= ~PCI_EXP_LNKCTL_LD;
 	err = pci_write_config_word(bridge, cap + PCI_EXP_LNKCTL, reg16);
 	if (err)
-		return err;
+		return pcibios_err_to_errno(err);
 
 	/* Check link */
 	if (!bridge->link_active_reporting) {
@@ -408,7 +408,7 @@ static int mlx5_pci_link_toggle(struct mlx5_core_dev *dev)
 	do {
 		err = pci_read_config_word(bridge, cap + PCI_EXP_LNKSTA, &reg16);
 		if (err)
-			return err;
+			return pcibios_err_to_errno(err);
 		if (reg16 & PCI_EXP_LNKSTA_DLLLA)
 			break;
 		msleep(20);

drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c static int mlx5_pci_link_toggle(struct mlx5_core_dev *dev) · line 432

Quoted 8 lines
drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c +1 -1
@@ -426,7 +426,7 @@ static int mlx5_pci_link_toggle(struct mlx5_core_dev *dev)
 	do {
 		err = pci_read_config_word(dev->pdev, PCI_DEVICE_ID, &reg16);
 		if (err)
-			return err;
+			return pcibios_err_to_errno(err);
 		if (reg16 == dev_id)
 			break;
 		msleep(20);

commit a48c6af2d2a5 ("net/mlx5: Convert PCI error values to generic errnos")

Author: Ilpo Järvinen <[email protected]>
Date:   Mon Aug 14 16:27:20 2023 +0300

net/mlx5: Convert PCI error values to generic errnos

mlx5_pci_link_toggle() returns a mix of PCI-specific error codes and generic errnos.

Convert the PCI-specific error values to generic errno using pcibios_err_to_errno() before returning them.

Fixes eabe8e5e88f5 ("net/mlx5: Handle sync reset now event")
Fixes 212b4d7251c1 ("net/mlx5: Wait for firmware to enable CRS before pci_restore_state")
[bhelgaas: rebase to pci/pcie-rmw, also convert in mlx5_check_dev_ids()]
drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c +6 -6
index 99dcbd006357..85a2dfbb5c46 100644
@@ -311,7 +311,7 @@ static int mlx5_check_dev_ids(struct mlx5_core_dev *dev, u16 dev_id)
 	list_for_each_entry(sdev, &bridge_bus->devices, bus_list) {
 		err = pci_read_config_word(sdev, PCI_DEVICE_ID, &sdev_id);
 		if (err)
-			return err;
+			return pcibios_err_to_errno(err);
 		if (sdev_id != dev_id) {
 			mlx5_core_warn(dev, "unrecognized dev_id (0x%x)\n", sdev_id);
 			return -EPERM;
@@ -371,7 +371,7 @@ static int mlx5_pci_link_toggle(struct mlx5_core_dev *dev)
 
 	err = pci_read_config_word(dev->pdev, PCI_DEVICE_ID, &dev_id);
 	if (err)
-		return err;
+		return pcibios_err_to_errno(err);
 	err = mlx5_check_dev_ids(dev, dev_id);
 	if (err)
 		return err;
@@ -386,11 +386,11 @@ static int mlx5_pci_link_toggle(struct mlx5_core_dev *dev)
 	/* PCI link toggle */
 	err = pcie_capability_set_word(bridge, PCI_EXP_LNKCTL, PCI_EXP_LNKCTL_LD);
 	if (err)
-		return err;
+		return pcibios_err_to_errno(err);
 	msleep(500);
 	err = pcie_capability_clear_word(bridge, PCI_EXP_LNKCTL, PCI_EXP_LNKCTL_LD);
 	if (err)
-		return err;
+		return pcibios_err_to_errno(err);
 
 	/* Check link */
 	if (!bridge->link_active_reporting) {
@@ -403,7 +403,7 @@ static int mlx5_pci_link_toggle(struct mlx5_core_dev *dev)
 	do {
 		err = pci_read_config_word(bridge, cap + PCI_EXP_LNKSTA, &reg16);
 		if (err)
-			return err;
+			return pcibios_err_to_errno(err);
 		if (reg16 & PCI_EXP_LNKSTA_DLLLA)
 			break;
 		msleep(20);
@@ -421,7 +421,7 @@ static int mlx5_pci_link_toggle(struct mlx5_core_dev *dev)
 	do {
 		err = pci_read_config_word(dev->pdev, PCI_DEVICE_ID, &reg16);
 		if (err)
-			return err;
+			return pcibios_err_to_errno(err);
 		if (reg16 == dev_id)
 			break;
 		msleep(20);
#2 lore raw
Quoted 11 lines · Bjorn Helgaas
Quoted 10 lines · Ilpo Järvinen

mlx5_pci_link_toggle() returns mix PCI specific error codes and generic errnos.

Convert the PCI specific error values to generic errno using pcibios_err_to_errno() before returning them.

Fixes eabe8e5e88f5 ("net/mlx5: Handle sync reset now event")
Fixes 212b4d7251c1 ("net/mlx5: Wait for firmware to enable CRS before pci_restore_state")
Quoted 8 lines
Quoted 5 lines

---

Maintainers beware, this will conflict with read+write -> set/clear_word fixes in pci.git/pcie-rmw. As such, it might be the easiest for Bjorn to take it instead of net people.

I provisionally rebased and applied it on pci/pcie-rmw. Take a look and make sure I didn't botch it --

Looks okay.

Quoted 2 lines

I also found a case in
mlx5_check_dev_ids() that looks like it needs the same conversion.

Ah, that where the one of them went (my first version had that fixed inside link_toggle but then when rebasing I didn't realize it had moved into another function).

Quoted 10 lines

The commit as applied is below.

If networking folks would prefer to take this, let me know and I can drop it.

Quoted 4 lines

I wonder if these PCIBIOS_* error codes are useful at all? There's 1:1 mapping into errno values so no information loss if the functions would just return errnos directly. Perhaps this is just legacy nobody has bothered to remove? If nobody opposes, I could take a look at getting rid of them.

Quoted 5 lines

I don't think the PCIBIOS error codes are very useful outside of arch/x86. They're returned by x86 PCIBIOS functions, and I think we still use those calls, but I don't think there's value in exposing the x86 error codes outside arch/x86. Looks like a big job to clean it up though ;)

Hmm... Do you mean pci_bios_read/write() in arch/x86/pci/pcbios.c? ...Because those functions are already inconsistent even with themselves, returning either -EINVAL or the PCI BIOS error code (or what I assume that masking of result to yield).

And unfortunately, that's far from the only inconsistency within arch PCI read/write func return values...

Signature · 1 line
 i.
#3 lore raw
Quoted 18 lines · Ilpo Järvinen
Quoted 11 lines · Bjorn Helgaas
Quoted 9 lines · Ilpo Järvinen

mlx5_pci_link_toggle() returns mix PCI specific error codes and generic errnos.
...

I wonder if these PCIBIOS_* error codes are useful at all? There's 1:1 mapping into errno values so no information loss if the functions would just return errnos directly. Perhaps this is just legacy nobody has bothered to remove? If nobody opposes, I could take a look at getting rid of them.

Quoted 5 lines

I don't think the PCIBIOS error codes are very useful outside of arch/x86. They're returned by x86 PCIBIOS functions, and I think we still use those calls, but I don't think there's value in exposing the x86 error codes outside arch/x86. Looks like a big job to clean it up though ;)

Quoted 4 lines

Hmm... Do you mean pci_bios_read/write() in arch/x86/pci/pcbios.c? ...Because those functions are already inconsistent even with themselves, returning either -EINVAL or the PCI BIOS error code (or what I assume that masking of result to yield).

I didn't look up the code; I just think we still use those PCIBIOS calls in some cases, so we need to know how to interpret the error values returned by the BIOS. IMHO it would be ideal if those PCIBIOS errors got converted to Linux errnos before generic code saw them.

Bjorn

#4 lore raw
In reply to Bjorn Helgaas
Quoted 17 lines · Bjorn Helgaas
Quoted 10 lines · Ilpo Järvinen

mlx5_pci_link_toggle() returns mix PCI specific error codes and generic errnos.

Convert the PCI specific error values to generic errno using pcibios_err_to_errno() before returning them.

Fixes eabe8e5e88f5 ("net/mlx5: Handle sync reset now event")
Fixes 212b4d7251c1 ("net/mlx5: Wait for firmware to enable CRS before pci_restore_state")
Quoted 5 lines

---

Maintainers beware, this will conflict with read+write -> set/clear_word fixes in pci.git/pcie-rmw. As such, it might be the easiest for Bjorn to take it instead of net people.

Quoted 9 lines

I provisionally rebased and applied it on pci/pcie-rmw. Take a look and make sure I didn't botch it -- I also found a case in mlx5_check_dev_ids() that looks like it needs the same conversion.

The commit as applied is below.

If networking folks would prefer to take this, let me know and I can drop it.

I Just took this patch into my mlx5 submission queue and sent it to netdev tree, please drop it from your tree.

Thanks for the patch,
Saeed.

#5 lore raw
Quoted 22 lines · Saeed Mahameed
Quoted 21 lines · Bjorn Helgaas
Quoted 10 lines · Ilpo Järvinen

mlx5_pci_link_toggle() returns mix PCI specific error codes and generic errnos.

Convert the PCI specific error values to generic errno using pcibios_err_to_errno() before returning them.

Fixes eabe8e5e88f5 ("net/mlx5: Handle sync reset now event")
Fixes 212b4d7251c1 ("net/mlx5: Wait for firmware to enable CRS before pci_restore_state")
Quoted 5 lines

---

Maintainers beware, this will conflict with read+write -> set/clear_word fixes in pci.git/pcie-rmw. As such, it might be the easiest for Bjorn to take it instead of net people.

I provisionally rebased and applied it on pci/pcie-rmw. Take a look and make sure I didn't botch it -- I also found a case in mlx5_check_dev_ids() that looks like it needs the same conversion.

Quoted 7 lines
Quoted 4 lines

The commit as applied is below.

If networking folks would prefer to take this, let me know and I can drop it.

I Just took this patch into my mlx5 submission queue and sent it to netdev tree, please drop it from your tree.

OK, will do. Note that this will generate a conflict between netdev and the PCI tree during the merge window.

Bjorn

maintainer
#6 lore raw
Quoted 25 lines · Bjorn Helgaas
Quoted 24 lines · Saeed Mahameed
Quoted 23 lines · Bjorn Helgaas
Quoted 10 lines · Ilpo Järvinen

mlx5_pci_link_toggle() returns mix PCI specific error codes and generic errnos.

Convert the PCI specific error values to generic errno using pcibios_err_to_errno() before returning them.

Fixes eabe8e5e88f5 ("net/mlx5: Handle sync reset now event")
Fixes 212b4d7251c1 ("net/mlx5: Wait for firmware to enable CRS before pci_restore_state")
Quoted 5 lines

---

Maintainers beware, this will conflict with read+write -> set/clear_word fixes in pci.git/pcie-rmw. As such, it might be the easiest for Bjorn to take it instead of net people.

I provisionally rebased and applied it on pci/pcie-rmw. Take a look and make sure I didn't botch it -- I also found a case in mlx5_check_dev_ids() that looks like it needs the same conversion.

The commit as applied is below.

Quoted 9 lines
Quoted 5 lines
Quoted 2 lines

If networking folks would prefer to take this, let me know and I can drop it.

I Just took this patch into my mlx5 submission queue and sent it to netdev tree, please drop it from your tree.

OK, will do. Note that this will generate a conflict between netdev and the PCI tree during the merge window.

In such case let me drop it and you submit it, I was worried it will conflict with another ongoing feature in mlx5, but I am almost sure it won't be ready this cycle, so no reason to panic, you can take the patch to the pci tree and I will revise my PR, it's not too late.

Quoted 1 line

Bjorn

#7 lore raw
Quoted 32 lines · Saeed Mahameed
Quoted 31 lines · Bjorn Helgaas
Quoted 29 lines · Saeed Mahameed
Quoted 17 lines · Bjorn Helgaas
Quoted 10 lines · Ilpo Järvinen

mlx5_pci_link_toggle() returns mix PCI specific error codes and generic errnos.

Convert the PCI specific error values to generic errno using pcibios_err_to_errno() before returning them.

Fixes eabe8e5e88f5 ("net/mlx5: Handle sync reset now event")
Fixes 212b4d7251c1 ("net/mlx5: Wait for firmware to enable CRS before pci_restore_state")
Quoted 5 lines

---

Maintainers beware, this will conflict with read+write -> set/clear_word fixes in pci.git/pcie-rmw. As such, it might be the easiest for Bjorn to take it instead of net people.

Quoted 8 lines

I provisionally rebased and applied it on pci/pcie-rmw. Take a look and make sure I didn't botch it -- I also found a case in mlx5_check_dev_ids() that looks like it needs the same conversion.

The commit as applied is below.

If networking folks would prefer to take this, let me know and I can drop it.

I Just took this patch into my mlx5 submission queue and sent it to netdev tree, please drop it from your tree.

Quoted 8 lines
Quoted 3 lines

OK, will do. Note that this will generate a conflict between netdev and the PCI tree during the merge window.

In such case let me drop it and you submit it, I was worried it will conflict with another ongoing feature in mlx5, but I am almost sure it won't be ready this cycle, so no reason to panic, you can take the patch to the pci tree and I will revise my PR, it's not too late.

OK, I'll keep it, I think that will be easier all around.

Bjorn