MPC5200, PCI, udelay(10)

3 messages, 3 authors, 2006-02-04 · open the first message on its own page

MPC5200, PCI, udelay(10)

From: Roman Kuzmenko <hidden>
Date: 2006-01-20 08:31:44

Hello.

Please, help me to investigate why the delays present in the
"mpc5xxx_read_config_dword" and "mpc5xxx_write_config_dword" functions
in the "arch/ppc/kernel/mpc52xx_pci.c" file.

I.e.
==========================
  *(volatile u32 *)MPC5xxx_PCI_CAR = 0;
udelay(10);
  return 0;
==========================

I was neither able to find any reference at the MPC5200 documentation
supplied by FreeScale nor reproduce any error with these lines got
removed.

Thank you.

-- 
Roman Kuzmenko [off-list ref]
Auriga, Inc.

Re: MPC5200, PCI, udelay(10)

From: Andrey Volkov <hidden>
Date: 2006-01-20 17:24:58

Roman Kuzmenko wrote:
Hello.

Please, help me to investigate why the delays present in the
"mpc5xxx_read_config_dword" and "mpc5xxx_write_config_dword" functions
in the "arch/ppc/kernel/mpc52xx_pci.c" file.

I.e.
==========================
  *(volatile u32 *)MPC5xxx_PCI_CAR = 0;
quoted
udelay(10);
  return 0;
==========================

I was neither able to find any reference at the MPC5200 documentation
supplied by FreeScale nor reproduce any error with these lines got
removed.

Thank you.
kernel version?

--
Regards
Andrey Volkov

Re: MPC5200, PCI, udelay(10)

From: Sylvain Munaut <hidden>
Date: 2006-02-04 10:51:14

Roman Kuzmenko wrote:
Hello.

Please, help me to investigate why the delays present in the
"mpc5xxx_read_config_dword" and "mpc5xxx_write_config_dword" functions
in the "arch/ppc/kernel/mpc52xx_pci.c" file.

I.e.
==========================
  *(volatile u32 *)MPC5xxx_PCI_CAR = 0;
quoted
udelay(10);
  return 0;
==========================

I was neither able to find any reference at the MPC5200 documentation
supplied by FreeScale nor reproduce any error with these lines got
removed.
I wasn't able to reproduce any problem either but people kept reporting
problems. But it seems that it's not a delay that's needed but a mb() to
ensure the write all _completly_ done before the next write. A
udelay(10) almost ensure that also but is a lot less elegant imho.


	Sylvain
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help