mmap /dev/mem in python

mmap /dev/mem in python

[Brett Graham]

I'm trying to mmap /dev/mem in python to access various registers on a
gumstix overo (with an omap 3530 processor). Here is the code I'm
using (as a minimal example):
----------------------------
import os, mmap
MAP_MASK = mmap.PAGESIZE - 1
addr = (see below)

f = os.open("/dev/mem", os.O_RDWR | os.O_SYNC)
m = mmap.mmap(f, mmap.PAGESIZE, mmap.MAP_SHARED, mmap.PROT_WRITE |
mmap.PROT_READ, offset=addr & ~MAP_MASK)
m.seek(addr & MAP_MASK)
c = m.read_byte()
print c
m.close()
os.close(f)
----------------------------

If I use: addr = 0x48002178 everything works swimmingly. However, if I
use: addr = 0x48088024 (or all the addresses I've tried > 0x48044000)
I get the following error

Unhandled fault: external abort on non-linefetch (0x1018) at 0x40020024
Bus error

and python crashes. I'm basing the mmap flags and options off of
devmem2.c. If I remove the O_SYNC flag from the os.open command, I can
read and write to registers fine, but randomly I will get the
Unhandled fault error and things will not work correctly.

Am I using the right command to write to the omap registers? should I
be using the O_SYNC flag? Any ideas where these errors are coming
from?

Thanks in advance for your help.

Re: mmap /dev/mem in python

[Tony Lindgren]

* Brett Graham <graham@rowland.harvard.edu> [091020 15:06]:
> I'm trying to mmap /dev/mem in python to access various registers on a
> gumstix overo (with an omap 3530 processor). Here is the code I'm
> using (as a minimal example):
> ----------------------------
> import os, mmap
> MAP_MASK = mmap.PAGESIZE - 1
> addr = (see below)
> 
> f = os.open("/dev/mem", os.O_RDWR | os.O_SYNC)
> m = mmap.mmap(f, mmap.PAGESIZE, mmap.MAP_SHARED, mmap.PROT_WRITE |
> mmap.PROT_READ, offset=addr & ~MAP_MASK)
> m.seek(addr & MAP_MASK)
> c = m.read_byte()
> print c
> m.close()
> os.close(f)
> ----------------------------
> 
> If I use: addr = 0x48002178 everything works swimmingly. However, if I
> use: addr = 0x48088024 (or all the addresses I've tried > 0x48044000)
> I get the following error
> 
> Unhandled fault: external abort on non-linefetch (0x1018) at 0x40020024
> Bus error
> 
> and python crashes. I'm basing the mmap flags and options off of
> devmem2.c. If I remove the O_SYNC flag from the os.open command, I can
> read and write to registers fine, but randomly I will get the
> Unhandled fault error and things will not work correctly.
> 
> Am I using the right command to write to the omap registers? should I
> be using the O_SYNC flag? Any ideas where these errors are coming
> from?

Sounds like you need to enable ick and fck for the device you're trying
to read.

Tony

Fwd: mmap /dev/mem in python

[Brett Graham]

If I run the following commads:
devmem2 0x48004a40 w 0x0000038a # select 32kHz clock
devmem2 0x48004a10 w 0x03007052 # enable iclk for gpt11
devmem2 0x48004a00 w 0x03007000 # enable fclk for gpt11

which all run sensibly and alter the registers, then I try to previous
minimum example on address 0x48088024 (TCLK for gpt11) I get the
following error:
Unhandled fault: external abort on non-linefetch (0x1018) at 0x40020024
Bus error

What puzzles me is that when I use devmem2 to read or write to the
same register, everything works fine.

On Tue, Oct 20, 2009 at 6:17 PM, Tony Lindgren <tony@atomide.com> wrote:
> * Brett Graham <graham@rowland.harvard.edu> [091020 15:06]:
>> I'm trying to mmap /dev/mem in python to access various registers on a
>> gumstix overo (with an omap 3530 processor). Here is the code I'm
>> using (as a minimal example):
>> ----------------------------
>> import os, mmap
>> MAP_MASK = mmap.PAGESIZE - 1
>> addr = (see below)
>>
>> f = os.open("/dev/mem", os.O_RDWR | os.O_SYNC)
>> m = mmap.mmap(f, mmap.PAGESIZE, mmap.MAP_SHARED, mmap.PROT_WRITE |
>> mmap.PROT_READ, offset=addr & ~MAP_MASK)
>> m.seek(addr & MAP_MASK)
>> c = m.read_byte()
>> print c
>> m.close()
>> os.close(f)
>> ----------------------------
>>
>> If I use: addr = 0x48002178 everything works swimmingly. However, if I
>> use: addr = 0x48088024 (or all the addresses I've tried > 0x48044000)
>> I get the following error
>>
>> Unhandled fault: external abort on non-linefetch (0x1018) at 0x40020024
>> Bus error
>>
>> and python crashes. I'm basing the mmap flags and options off of
>> devmem2.c. If I remove the O_SYNC flag from the os.open command, I can
>> read and write to registers fine, but randomly I will get the
>> Unhandled fault error and things will not work correctly.
>>
>> Am I using the right command to write to the omap registers? should I
>> be using the O_SYNC flag? Any ideas where these errors are coming
>> from?
>
> Sounds like you need to enable ick and fck for the device you're trying
> to read.
>
> Tony
> --
> To unsubscribe from this list: send the line "unsubscribe linux-omap" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: mmap /dev/mem in python

[Paul Walmsley]

On Tue, 20 Oct 2009, Brett Graham wrote:

> If I use: addr = 0x48002178 everything works swimmingly. However, if I
> use: addr = 0x48088024 (or all the addresses I've tried > 0x48044000)
> I get the following error
> 
> Unhandled fault: external abort on non-linefetch (0x1018) at 0x40020024
> Bus error

0x40020024 != 0x48088024 ?  That's probably not the kernel's fault.


- Paul

Re: Fwd: mmap /dev/mem in python

[Paul Walmsley]

On Tue, 20 Oct 2009, Brett Graham wrote:

> If I run the following commads:
> devmem2 0x48004a40 w 0x0000038a # select 32kHz clock
> devmem2 0x48004a10 w 0x03007052 # enable iclk for gpt11
> devmem2 0x48004a00 w 0x03007000 # enable fclk for gpt11
> 
> which all run sensibly and alter the registers, then I try to previous
> minimum example on address 0x48088024 (TCLK for gpt11) I get the
> following error:
> Unhandled fault: external abort on non-linefetch (0x1018) at 0x40020024
> Bus error
> 
> What puzzles me is that when I use devmem2 to read or write to the
> same register, everything works fine.

Consider running strace on devmem2 and comparing it to the strace of your 
Python code?  That should show you any differences.  If devmem2 works, 
it's unlikely to be a kernel problem.


- Paul

Re: Fwd: mmap /dev/mem in python

[Brett Graham]

It's good to at least know that this doesn't seem to be a kernel
issue, and thanks for the strace advice (I am new to this level of
linux development). Here is the truncated output for devmem and the
minimal example:

--- devmem ---
open("/dev/mem", O_RDWR|O_SYNC)         = 3
fstat64(1, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 0), ...}) = 0
mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1,
0) = 0x4001f000
write(1, "/dev/mem opened.\n", 17)      = 17
mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_SHARED, 3, 0x48088) = 0x40020000
write(1, "Memory mapped at address 0x40020"..., 37) = 37
write(1, "Value at address 0x48088024 (0x4"..., 46) = 46
munmap(0x40020000, 4096)                = 0
close(3)                                = 0
io_submit(0, 0, 0xfbad2088 <unfinished ... exit status 0>
Process 1635 detached


--minimal python example--
open("/dev/mem", O_RDWR|O_SYNC|O_LARGEFILE) = 3
fstat64(3, {st_mode=S_IFCHR|0640, st_rdev=makedev(1, 1), ...}) = 0
dup(3)                                  = 4
mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_SHARED, 3, 0x48088) = 0x40020000
--- SIGBUS (Bus error) @ 0 (0) ---
+++ killed by SIGBUS +++
Process 1633 detached

It looks like python decided to add a O_LARGEFILE and duplicate the
file descriptor. I'm still not sure what's going on here that's
causing the problem as both mmap2 calls are identical.

I don't mean to bombard this list with a problem that isn't related to
the kernel, so I'll post this to a python list somewhere. Thanks a
million for your help so far, this is driving me nutty! Also if any
more info would help just let me know.

On Tue, Oct 20, 2009 at 7:26 PM, Paul Walmsley <paul@pwsan.com> wrote:
> On Tue, 20 Oct 2009, Brett Graham wrote:
>
>> If I run the following commads:
>> devmem2 0x48004a40 w 0x0000038a # select 32kHz clock
>> devmem2 0x48004a10 w 0x03007052 # enable iclk for gpt11
>> devmem2 0x48004a00 w 0x03007000 # enable fclk for gpt11
>>
>> which all run sensibly and alter the registers, then I try to previous
>> minimum example on address 0x48088024 (TCLK for gpt11) I get the
>> following error:
>> Unhandled fault: external abort on non-linefetch (0x1018) at 0x40020024
>> Bus error
>>
>> What puzzles me is that when I use devmem2 to read or write to the
>> same register, everything works fine.
>
> Consider running strace on devmem2 and comparing it to the strace of your
> Python code?  That should show you any differences.  If devmem2 works,
> it's unlikely to be a kernel problem.
>
>
> - Paul
>
--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Fwd: mmap /dev/mem in python

[Paul Walmsley]

On Tue, 20 Oct 2009, Brett Graham wrote:

> --minimal python example--
> open("/dev/mem", O_RDWR|O_SYNC|O_LARGEFILE) = 3
> fstat64(3, {st_mode=S_IFCHR|0640, st_rdev=makedev(1, 1), ...}) = 0
> dup(3)                                  = 4
> mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_SHARED, 3, 0x48088) = 0x40020000
> --- SIGBUS (Bus error) @ 0 (0) ---
> +++ killed by SIGBUS +++
> Process 1633 detached

Your script:

>> m = mmap.mmap(f, mmap.PAGESIZE, mmap.MAP_SHARED, mmap.PROT_WRITE |
>> mmap.PROT_READ, offset=addr & ~MAP_MASK)
>> m.seek(addr & MAP_MASK)

So the mmap() is working fine, based on the output above...

>> c = m.read_byte()

... but the above line is going to fail.  GPTIMER registers need 16- 
or 32-bit accesses[1].


- Paul

1. OMAP34xx TRM 16.3.2:

"CAUTION 
The GP timer registers are limited to 32-bit and 16-bit data accesses; 
8-bit access is not allowed and can corrupt the register content."

Re: Fwd: mmap /dev/mem in python

[Brett Graham]

Doh! Thank you! This looks to be the problem, I've made some changes
to the code (calling mmap.read(4) rather than mmap.read_bye()) and
everything seems to be working.

On Tue, Oct 20, 2009 at 8:27 PM, Paul Walmsley <paul@pwsan.com> wrote:
> On Tue, 20 Oct 2009, Brett Graham wrote:
>
>> --minimal python example--
>> open("/dev/mem", O_RDWR|O_SYNC|O_LARGEFILE) = 3
>> fstat64(3, {st_mode=S_IFCHR|0640, st_rdev=makedev(1, 1), ...}) = 0
>> dup(3)                                  = 4
>> mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_SHARED, 3, 0x48088) = 0x40020000
>> --- SIGBUS (Bus error) @ 0 (0) ---
>> +++ killed by SIGBUS +++
>> Process 1633 detached
>
> Your script:
>
>>> m = mmap.mmap(f, mmap.PAGESIZE, mmap.MAP_SHARED, mmap.PROT_WRITE |
>>> mmap.PROT_READ, offset=addr & ~MAP_MASK)
>>> m.seek(addr & MAP_MASK)
>
> So the mmap() is working fine, based on the output above...
>
>>> c = m.read_byte()
>
> ... but the above line is going to fail.  GPTIMER registers need 16-
> or 32-bit accesses[1].
>
>
> - Paul
>
> 1. OMAP34xx TRM 16.3.2:
>
> "CAUTION
> The GP timer registers are limited to 32-bit and 16-bit data accesses;
> 8-bit access is not allowed and can corrupt the register content."
>
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html