Skip to contentSkip to conversation

PCI: kirin: Fix section mismatch warning

PATCH Superseded Nathan Chancellor 3 participants 6 messages Sep 19, 2018

Subsystems (3)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 PCI: kirin: Fix section mismatch warning

WARNING: vmlinux.o(.text+0x4758cc): Section mismatch in reference from the function kirin_pcie_probe() to the function .init.text:kirin_add_pcie_port()
The function kirin_pcie_probe() references
the function __init kirin_add_pcie_port().
This is often because kirin_pcie_probe lacks a __init annotation or the annotation of kirin_add_pcie_port is wrong.

Remove '__init' from kirin_add_pcie_port so there is no mismatch.

Fixes fc5165db245a ("PCI: kirin: Add HiSilicon Kirin SoC PCIe controller driver")
---
 drivers/pci/controller/dwc/pcie-kirin.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/pci/controller/dwc/pcie-kirin.c +2 -2
index 5352e0c3be82..9b599296205d 100644
@@ -467,8 +467,8 @@ static int kirin_pcie_add_msi(struct dw_pcie *pci,
 	return 0;
 }
 
-static int __init kirin_add_pcie_port(struct dw_pcie *pci,
-				      struct platform_device *pdev)
+static int kirin_add_pcie_port(struct dw_pcie *pci,
+			       struct platform_device *pdev)
 {
 	int ret;
 
Signature · 1 line
2.19.0
#1 lore raw

Xiaowei, Binghui,

Quoted 9 lines · Nathan Chancellor

WARNING: vmlinux.o(.text+0x4758cc): Section mismatch in reference from the function kirin_pcie_probe() to the function .init.text:kirin_add_pcie_port()
The function kirin_pcie_probe() references
the function __init kirin_add_pcie_port().
This is often because kirin_pcie_probe lacks a __init annotation or the annotation of kirin_add_pcie_port is wrong.

Remove '__init' from kirin_add_pcie_port so there is no mismatch.

I think that instead of removing __init section tag we should add it to kirin_pcie_probe().

Please let me know and ACK accordingly.

Lorenzo

Quoted 25 lines
Fixes fc5165db245a ("PCI: kirin: Add HiSilicon Kirin SoC PCIe controller driver")
---
 drivers/pci/controller/dwc/pcie-kirin.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/pci/controller/dwc/pcie-kirin.c +2 -2
index 5352e0c3be82..9b599296205d 100644
@@ -467,8 +467,8 @@ static int kirin_pcie_add_msi(struct dw_pcie *pci,
 	return 0;
 }
 
-static int __init kirin_add_pcie_port(struct dw_pcie *pci,
-				      struct platform_device *pdev)
+static int kirin_add_pcie_port(struct dw_pcie *pci,
+			       struct platform_device *pdev)
 {
 	int ret;
 
Signature · 1 line
2.19.0
#2 lore raw
Quoted 12 lines · Lorenzo Pieralisi

Xiaowei, Binghui,

Quoted 8 lines · Nathan Chancellor

WARNING: vmlinux.o(.text+0x4758cc): Section mismatch in reference from the function kirin_pcie_probe() to the function .init.text:kirin_add_pcie_port()
The function kirin_pcie_probe() references
the function __init kirin_add_pcie_port().
This is often because kirin_pcie_probe lacks a __init annotation or the annotation of kirin_add_pcie_port is wrong.

Quoted 4 lines
Quoted 1 line

Remove '__init' from kirin_add_pcie_port so there is no mismatch.

I think that instead of removing __init section tag we should add it to kirin_pcie_probe().

A bunch of these functions have a single call site, with a mix of __init:

kirin_pcie_probe() ->
kirin_add_pcie_port() -> // __init
kirin_pcie_add_msi()

But it looks like probe functions should be marked init:

$ cd drivers; grep -r __init | grep probe

seems to turn up quite a lot of *probe() functions that are marked __init.
https://www.tldp.org/LDP/lkmpg/2.4/html/x281.htm has more info, and from reading that, it seems that it's a no-op for a module that's marked loadable.

So I agree; __init should be added to kirin_pcie_probe(), but also kirin_pcie_add_msi(). That way if the module is statically compiled into the kernel image, these functions with a single call site get cleaned up after init.

Nathan, would you mind sending a v2?

Quoted 30 lines

Please let me know and ACK accordingly.

Lorenzo

Quoted 25 lines
Fixes fc5165db245a ("PCI: kirin: Add HiSilicon Kirin SoC PCIe controller driver")
---
 drivers/pci/controller/dwc/pcie-kirin.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/pci/controller/dwc/pcie-kirin.c +2 -3
index 5352e0c3be82..9b599296205d 100644
@@ -467,8 +467,8 @@ static int kirin_pcie_add_msi(struct dw_pcie *pci,
      return 0;
 }
 
-static int __init kirin_add_pcie_port(struct dw_pcie *pci,
-                                   struct platform_device *pdev)
+static int kirin_add_pcie_port(struct dw_pcie *pci,
+                            struct platform_device *pdev)
 {
      int ret;
 
--

2.19.0

Signature · 2 lines
Thanks,
~Nick Desaulniers
#3 lore raw
Quoted 33 lines · Nick Desaulniers
Quoted 12 lines · Lorenzo Pieralisi

Xiaowei, Binghui,

Quoted 8 lines · Nathan Chancellor

WARNING: vmlinux.o(.text+0x4758cc): Section mismatch in reference from the function kirin_pcie_probe() to the function .init.text:kirin_add_pcie_port()
The function kirin_pcie_probe() references
the function __init kirin_add_pcie_port().
This is often because kirin_pcie_probe lacks a __init annotation or the annotation of kirin_add_pcie_port is wrong.

Quoted 4 lines
Quoted 1 line

Remove '__init' from kirin_add_pcie_port so there is no mismatch.

I think that instead of removing __init section tag we should add it to kirin_pcie_probe().

A bunch of these functions have a single call site, with a mix of __init:

kirin_pcie_probe() ->
kirin_add_pcie_port() -> // __init
kirin_pcie_add_msi()

But it looks like probe functions should be marked init:

$ cd drivers; grep -r __init | grep probe

seems to turn up quite a lot of *probe() functions that are marked __init.
https://www.tldp.org/LDP/lkmpg/2.4/html/x281.htm has more info, and from reading that, it seems that it's a no-op for a module that's marked loadable.

Quoted 7 lines

So I agree; __init should be added to kirin_pcie_probe(), but also kirin_pcie_add_msi(). That way if the module is statically compiled into the kernel image, these functions with a single call site get cleaned up after init.

Nathan, would you mind sending a v2?

Hi Nick and Lorenzo,

I had tested adding __init to both kirin_pcie_probe and kirin_pcie_add_msi but ran into this warning:

WARNING: vmlinux.o(.data+0x68b78): Section mismatch in reference from the variable kirin_pcie_driver to the function .init.text:kirin_pcie_probe()
The variable kirin_pcie_driver references
the function __init kirin_pcie_probe()

Doing a quick 'grep -r __init | grep platform_driver' doesn't show any 'static struct platform_driver' constructs marked as __init, leading me to believe this patch is the proper solution.

I'm happy to hear otherwise, thank you both for the quick responses, Nathan

Quoted 36 lines
Quoted 30 lines

Please let me know and ACK accordingly.

Lorenzo

Quoted 25 lines
Fixes fc5165db245a ("PCI: kirin: Add HiSilicon Kirin SoC PCIe controller driver")
---
 drivers/pci/controller/dwc/pcie-kirin.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/pci/controller/dwc/pcie-kirin.c +2 -3
index 5352e0c3be82..9b599296205d 100644
@@ -467,8 +467,8 @@ static int kirin_pcie_add_msi(struct dw_pcie *pci,
      return 0;
 }
 
-static int __init kirin_add_pcie_port(struct dw_pcie *pci,
-                                   struct platform_device *pdev)
+static int kirin_add_pcie_port(struct dw_pcie *pci,
+                            struct platform_device *pdev)
 {
      int ret;
 
--

2.19.0

Signature · 2 lines
Thanks,
~Nick Desaulniers
#4 lore raw
Quoted 43 lines · Nathan Chancellor
Quoted 33 lines · Nick Desaulniers
Quoted 12 lines · Lorenzo Pieralisi

Xiaowei, Binghui,

Quoted 8 lines · Nathan Chancellor

WARNING: vmlinux.o(.text+0x4758cc): Section mismatch in reference from the function kirin_pcie_probe() to the function .init.text:kirin_add_pcie_port()
The function kirin_pcie_probe() references
the function __init kirin_add_pcie_port().
This is often because kirin_pcie_probe lacks a __init annotation or the annotation of kirin_add_pcie_port is wrong.

Quoted 4 lines
Quoted 1 line

Remove '__init' from kirin_add_pcie_port so there is no mismatch.

I think that instead of removing __init section tag we should add it to kirin_pcie_probe().

A bunch of these functions have a single call site, with a mix of __init:

kirin_pcie_probe() ->
kirin_add_pcie_port() -> // __init
kirin_pcie_add_msi()

But it looks like probe functions should be marked init:

$ cd drivers; grep -r __init | grep probe

seems to turn up quite a lot of *probe() functions that are marked __init.
https://www.tldp.org/LDP/lkmpg/2.4/html/x281.htm has more info, and from reading that, it seems that it's a no-op for a module that's marked loadable.

Quoted 7 lines

So I agree; __init should be added to kirin_pcie_probe(), but also kirin_pcie_add_msi(). That way if the module is statically compiled into the kernel image, these functions with a single call site get cleaned up after init.

Nathan, would you mind sending a v2?

Quoted 7 lines

Hi Nick and Lorenzo,

I had tested adding __init to both kirin_pcie_probe and kirin_pcie_add_msi but ran into this warning:

WARNING: vmlinux.o(.data+0x68b78): Section mismatch in reference from the variable kirin_pcie_driver to the function

I think we want builtin_platform_driver_probe() here.

Try removing the probe reference from the declaration of kirin_pcie_driver, and then calling

builtin_platform_driver_probe(kirin_pcie_driver, kirin_pcie_probe);

Does that fix the section mismatches, while keeping as much in __init as possible?

Quoted 47 lines

.init.text:kirin_pcie_probe()
The variable kirin_pcie_driver references
the function __init kirin_pcie_probe()

Doing a quick 'grep -r __init | grep platform_driver' doesn't show any 'static struct platform_driver' constructs marked as __init, leading me to believe this patch is the proper solution.

I'm happy to hear otherwise, thank you both for the quick responses, Nathan

Quoted 36 lines
Quoted 27 lines

Please let me know and ACK accordingly.

Lorenzo

Quoted 22 lines
Fixes fc5165db245a ("PCI: kirin: Add HiSilicon Kirin SoC PCIe controller driver")
---
 drivers/pci/controller/dwc/pcie-kirin.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/pci/controller/dwc/pcie-kirin.c +2 -2
index 5352e0c3be82..9b599296205d 100644
@@ -467,8 +467,8 @@ static int kirin_pcie_add_msi(struct dw_pcie *pci,
      return 0;
 }
 
-static int __init kirin_add_pcie_port(struct dw_pcie *pci,
-                                   struct platform_device *pdev)
+static int kirin_add_pcie_port(struct dw_pcie *pci,
+                            struct platform_device *pdev)
 {
      int ret;
 
Quoted 3 lines
Quoted 3 lines

--
2.19.0

--
Thanks,
~Nick Desaulniers

Signature · 2 lines
Thanks,
~Nick Desaulniers
#5 lore raw
Quoted 55 lines · Nick Desaulniers
Quoted 43 lines · Nathan Chancellor
Quoted 33 lines · Nick Desaulniers
Quoted 12 lines · Lorenzo Pieralisi

Xiaowei, Binghui,

Quoted 8 lines · Nathan Chancellor

WARNING: vmlinux.o(.text+0x4758cc): Section mismatch in reference from the function kirin_pcie_probe() to the function .init.text:kirin_add_pcie_port()
The function kirin_pcie_probe() references
the function __init kirin_add_pcie_port().
This is often because kirin_pcie_probe lacks a __init annotation or the annotation of kirin_add_pcie_port is wrong.

Quoted 4 lines
Quoted 1 line

Remove '__init' from kirin_add_pcie_port so there is no mismatch.

I think that instead of removing __init section tag we should add it to kirin_pcie_probe().

A bunch of these functions have a single call site, with a mix of __init:

kirin_pcie_probe() ->
kirin_add_pcie_port() -> // __init
kirin_pcie_add_msi()

But it looks like probe functions should be marked init:

$ cd drivers; grep -r __init | grep probe

seems to turn up quite a lot of *probe() functions that are marked __init.
https://www.tldp.org/LDP/lkmpg/2.4/html/x281.htm has more info, and from reading that, it seems that it's a no-op for a module that's marked loadable.

Quoted 7 lines

So I agree; __init should be added to kirin_pcie_probe(), but also kirin_pcie_add_msi(). That way if the module is statically compiled into the kernel image, these functions with a single call site get cleaned up after init.

Nathan, would you mind sending a v2?

Quoted 7 lines

Hi Nick and Lorenzo,

I had tested adding __init to both kirin_pcie_probe and kirin_pcie_add_msi but ran into this warning:

WARNING: vmlinux.o(.data+0x68b78): Section mismatch in reference from the variable kirin_pcie_driver to the function

I think we want builtin_platform_driver_probe() here.

Quoted 8 lines

Try removing the probe reference from the declaration of kirin_pcie_driver, and then calling

builtin_platform_driver_probe(kirin_pcie_driver, kirin_pcie_probe);

Does that fix the section mismatches, while keeping as much in __init as possible?

Hi Nick,

Thank you for that suggestion, it worked. I will send a v2 for review shortly.

Nathan

Quoted 53 lines
Quoted 47 lines

.init.text:kirin_pcie_probe()
The variable kirin_pcie_driver references
the function __init kirin_pcie_probe()

Doing a quick 'grep -r __init | grep platform_driver' doesn't show any 'static struct platform_driver' constructs marked as __init, leading me to believe this patch is the proper solution.

I'm happy to hear otherwise, thank you both for the quick responses, Nathan

Quoted 36 lines
Quoted 27 lines

Please let me know and ACK accordingly.

Lorenzo

Quoted 22 lines
Fixes fc5165db245a ("PCI: kirin: Add HiSilicon Kirin SoC PCIe controller driver")
---
 drivers/pci/controller/dwc/pcie-kirin.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/pci/controller/dwc/pcie-kirin.c +2 -2
index 5352e0c3be82..9b599296205d 100644
@@ -467,8 +467,8 @@ static int kirin_pcie_add_msi(struct dw_pcie *pci,
      return 0;
 }
 
-static int __init kirin_add_pcie_port(struct dw_pcie *pci,
-                                   struct platform_device *pdev)
+static int kirin_add_pcie_port(struct dw_pcie *pci,
+                            struct platform_device *pdev)
 {
      int ret;
 
Quoted 3 lines
Quoted 3 lines

--
2.19.0

--
Thanks,
~Nick Desaulniers

Signature · 2 lines
Thanks,
~Nick Desaulniers