[PATCH] mips/pci: Reduce stack frame usage

[PATCH] mips/pci: Reduce stack frame usage

[Keith Busch]

This patch removes creating a fake pci device in MIPS early config
access and instead just uses the pci bus to get the same functionality.
The struct pci_dev is too large to allocate on the stack, and was relying
on compiler optimizations to remove its usage.

Signed-off-by: Keith Busch <keith.busch@intel.com>
---

While I don't have any hardware to test this, the change should be
exactly the same as before, taking the direct route to the config read
instead of letting the compiler optimize this.

This patch is preparing to add surprise removed device handling to the
pci_read_config_*, which makes the compiler optimization that currently
removes the excessive stack usage impossible.

 arch/mips/txx9/generic/pci.c | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/arch/mips/txx9/generic/pci.c b/arch/mips/txx9/generic/pci.c
index 1f6bc9a..285d84e 100644
--- a/arch/mips/txx9/generic/pci.c
+++ b/arch/mips/txx9/generic/pci.c
@@ -29,12 +29,8 @@ static int __init
 early_read_config_word(struct pci_controller *hose,
 		       int top_bus, int bus, int devfn, int offset, u16 *value)
 {
-	struct pci_dev fake_dev;
 	struct pci_bus fake_bus;
 
-	fake_dev.bus = &fake_bus;
-	fake_dev.sysdata = hose;
-	fake_dev.devfn = devfn;
 	fake_bus.number = bus;
 	fake_bus.sysdata = hose;
 	fake_bus.ops = hose->pci_ops;
@@ -45,7 +41,7 @@ early_read_config_word(struct pci_controller *hose,
 	else
 		fake_bus.parent = NULL;
 
-	return pci_read_config_word(&fake_dev, offset, value);
+	return pci_bus_read_config_word(&fake_bus, devfn, offset, value);
 }
 
 int __init txx9_pci66_check(struct pci_controller *hose, int top_bus,
-- 
2.7.2

Re: [PATCH] mips/pci: Reduce stack frame usage

[Ralf Baechle]

On Tue, Sep 13, 2016 at 09:21:47AM -0600, Keith Busch wrote:

> This patch removes creating a fake pci device in MIPS early config
> access and instead just uses the pci bus to get the same functionality.
> The struct pci_dev is too large to allocate on the stack, and was relying
> on compiler optimizations to remove its usage.
> 
> Signed-off-by: Keith Busch <keith.busch@intel.com>
> ---
> 
> While I don't have any hardware to test this, the change should be
> exactly the same as before, taking the direct route to the config read
> instead of letting the compiler optimize this.
> 
> This patch is preparing to add surprise removed device handling to the
> pci_read_config_*, which makes the compiler optimization that currently
> removes the excessive stack usage impossible.

Looks ok so I've applied it.  Atsushi?

  Ralf