ARM Linux - Mailing Lists

Home + About + Documentation + Developer + Mailing Lists + Contacting

Mailing list questions:
1. How do I unsubscribe from the mailing lists.
General questions:
Patch Submission (via git/patches etc)

1. Mailing list questions:

1.1. How do I subscribe or unsubscribe from the mailing lists.

The mailing list software appends a footer to every message sent via the mailing lists:

-------------------------------------------------------------------
List admin: http://lists.armlinux.org.uk/mailman/listinfo/linux-arm
FAQ:        http://www.armlinux.org.uk/mailinglists/faq.php
Etiquette:  http://www.armlinux.org.uk/mailinglists/etiquette.php

If you point your web browser to the "list admin" address and scroll to the bottom, where you can enter your subscribed email address. Once this is done, you will be given instructions on how to unsubscribe or change your options.

2. General questions:

2.1. How do I compile a kernel for ARM?

[rmk]
Please see the kernel compilation instructions.

2.2. I am trying to run a Linux kernel 2.6 on my Intel IXP42x. I am using the kernel 2.6.x. This kernel does not include support for the integrated ethernet controllers of the IXP425. I have downloaded the AccessLibrary 1.x from Intel, but i have seen it only supports kernel 2.4. Is there any news about a new version of the AccessLibrary for 2.6 kernel?

[19 March 2004 - ds]
Unfortunately, the only people who can help you are Intel b/c they need to port the access library to 2.6 since the code is under a closed license and even if someone outside Intel ported it to 2.6, they could not share the code due to the propietary license.

[14 December 2004 - rmk]
When we say "the only people who can help you are Intel" we mean it. Please do not post questions about this to the ARM Linux mailing lists on the off chance. It just winds certain people up. We can't help you. We can't even suggest where you might get help from, other than Intel directly. Talk to your Intel representative, or if you don't know who that is, your manager. The best information we have is in the above FAQ entry.

The linux-arm* lists are for supporting Open Source only and not for random proprietary closed source products.

[10 September 2008 - rwhitby] As of 2.6.26, the kernel includes support for the integrated ethernet controllers of the IXP42x (see drivers/net/arm/ixp4xx_eth.c, arch/arm/mach-ixp4xx/ixp4xx_npe.c and arch/arm/mach-ixp4xx/ixp4xx_qmgr.c). You are still required to download the proprietary firmware microcode from Intel, which is loaded via hotplug.

2.3. I feel really confused about what's the difference with gcc soft-float support and the kernel NWFPE/FastFPE support?

[31 January 2004 - np]
Scenario 1 (hard-float): The compiler emits opcodes designed to be used with a hardware floating point coprocessor (FPU). The FPU usually has a set of extra registers for its use, and the compiler may as well pass floating point argument to functions through those registers. This is of course the best performing solution when a real hardware FPU is used.

Scenario 2 (soft-float): the compiler converts floating point operations into function calls and a special library is used to provide all functions performing the required operations, all in software with no FPU instructions at all. There is obviously no extra floating point registers available in this case, therefore all FP arguments to functions have to be passed through standard registers or on the stack. This is of course the best performing solution when no hardware FPU is available, given that the library implementing the FP operations is optimally coded.

Now, unfortunately, the default on ARM Linux has traditionally been set to have the compiler use hard-float, even if ARM Linux never ran on any ARM CPU with a real hardware FPU. The CPU is therefore raising the invalid instruction exception each time some FPU opcode is encountered. Then the kernel traps that exception, looks at the given FPU instruction and emulates it in software. But here not only the kernel must perform the FP operation, it must emulates the whole hardware FPU as well. That's what NWFPE or FastFPE are doing. This is obviously the worst performing arrangement that can be due to the exception trap and emulation overhead

2.4. Can I use the both hard & soft float at the same time or only choose one of them?

[31 January 2004 - np]
Depends. You can use both at the same time, but not in the same application. The problem as to do with the ABI incompatibility between soft-float and hard-float due to the different floating point argument passing conventions. So, if you decide to switch to using soft-float for some application, you MUST recompile ALL the libraries that application is going to use, including system libraries like the C library. If those libraries are dynamically linked libraries, you then must also recompile all the applications that share those same libraries. And then your application will run correctly only on systems with soft-float environments, unless you link it statically in which case it will run anywhere (even on a kernel with NWFPE configured in -- it will simply not be invoked). That's the main reason why mainstream ARM distributions are still reluctant to switch to soft-float because of the associated compatibility pain.

2.5. When I compile my 2.6.x kernel for ARM I get a compiler/assembler error. Any idea ?

[2 April 2004 - rmk]
When reporting these problems, always show the full build line. This can be obtained by re-running the make with an additional argument: V=1

2.6. When I trace kernel in head.S I found problems that may be caused by __turn_mmu_on. Any idea ?

[5 April 2004 - rmk]
Please read the comment in head.S:

/*
 * Enable the MMU.  This completely changes the structure of the visible
 * memory space.  You will not be able to trace execution through this.
 * If you have an enquiry about this, *please* check the linux-arm-kernel
 * mailing list archives BEFORE sending another post to the list.
 */

It's been asked soo many times that it really is a frequently asked question. The above comment is supposed to head off further questions, but alas... Hopefully this FAQ entry will do the job.

[11 January 2002 - rmk]
The number 1 problem is that people just do not realise what "enabling the MMU" means. It means that _all_ of the addressable memory _will_ change, and this normally means various regions you'd like to write to no longer exist - in other words, if you have debugging code present that doesn't cater for a complete change in the memory structure, then your debugging code is buggy.

(Side note: I've even had people insist that their debugging code is perfect, even after mentioning the above, and then we've gone through several mails, ended up having the debugging code posted, and it turns out to buggy, assuming that the LEDs at some random address are still accessible.)

The best advice is if you're using your own debugging code, remove it. Use the debugging printascii() stuff for your platform in arch/arm/kernel/debug-armv.S (enabled by CONFIG_DEBUG_LL) and drop a printascii() into printk() (kernel/printk.c) just after the vsprintf() call, and monitor the relevant serial port.

2.7. I'm working on a bootloader, where can I find ARM Linux kernel booting requirements ?

[7 June 2004 - rmk]
First of all, we don't need another boot loader. If you're planning on writing one from scratch, please don't. Use one of the existing boot loaders, such as blob or uboot.

For kernel booting requirements, please see the website:

2.8. How can I access /dev/mem ? - How can I map memory in user space ?

[10 July 2004 - rmk]
No need to reinvent the wheel, you will find a sample implementation of this type of thing in devmem2. See http://www.lartmaker.nl/lartware/port/devmem2.c for the source.

2.9. I see '$a', '$d' and '$t' symbols in the backtrace/ksymoops output/System.map/linker error messages. What are they and how can I get rid of them ?

[12 July 2004 - rmk]
These are "mapping symbols", which are used in the generation of some binary formats, added by binutils 2.15. Unfortunately, they leak into the symbol table, and as such interfere with the linkers ability to find the correct symbol to report for error messages. They also appear in the objdump --syms and nm outputs, and affect the operation of the kernel backtrace and ksymoops outputs.

In order to work around this problem, I recommend not using binutils 2.15 for the time being. Please use an earlier binutils version like 2.14 instead. However, please ensure that you have applied this fix.

2.10. Where can I find -rmk -pxa -* patches for the 2.6 kernel ?

[21 December 2005 - eb]
Since 2.6, all the ARM support is directly pushed in the mainstream kernel, so you shouldn't need any patch. Just get the official 2.6.x stable kernel from http://www.kernel.org/ or the latest development version from the git tree at http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=summary. Russell King's development git tree is synced every two days with Linus' tree so you only need to keep synced with Linus' one. As usual, you will find patches awaiting review in Russell King's patch system here: http://www.armlinux.org.uk/developer/patches/

2.11. I'm using this 2.4 kernel, and...

[27 July 2006 - erikm]
2.4 support has been abandoned. Don't ask questions about 2.4 on these lists, use the latest 2.6 kernel, and try to recreate the problem. If it still persists, then ask on the list.

2.12. I'm using this 2.6 kernel, and...

[27 July 2006 - erikm]
First make sure that you are running the latest stable kernel. Bugs in older kernels are a waste of everyone's time. Try to recreate your problem with the latest stable kernel and if it persists, ask on the list. The latest stable kernel releases can be found at http://www.kernel.org/kdist/finger_banner

2.13. Why does my debugger trap in __arch_clear_user?

[24 August 2006 - rmk]
This is normal. As the kernel operates a demand paged memory system, pages are faulted in via data aborts and prefetch aborts. However, some debuggers assume that data or prefetch aborts means that your program did something wrong, and halt program execution on these events. Therefore, you need to disable this behaviour in your debugger.

2.14. Known toochain issues.

[14 July 2009]
The CodeSourcery Sourcery G++ Lite 2007q3-51 toolchain (gcc version 4.2.1) is broken with -fomit-frame-pointer. Without frame pointer, gcc tries to be smart and optimise the tail call. However, it gets it terribly wrong.

3. General questions:

3.1. Where should patches be reviewed?

[rmk]
Patches should be reviewed on the mailing list prior to submission. This forfills two aspects - allowing people on the mailing list to review your work, and letting others know of your work.

3.2. What does rmk's "ok" mean?

[rmk]
It means that rmk is happy with the patch for it to be submitted. It does not mean you can add an 'Acked-by' tag for rmk to a patch going to the patch system. 'Signed-off-by' should also not be used - when patches are merged from the patch system, rmk's sign-off will be automatically added.

3.3. Which git commit should I base my tree on?

[rmk]
Linus' tree, or some previously submitted point. Never on rmk's constantly regenerated devel branch - because it is constantly regenerated.

3.4. How should commits in my git tree be reviewed?

[rmk]
A possible way to export patches for mailing list review is using git-format-patch. This generates a set of patch files, one per commit, which can then be emailed as separate attachments (one per email message).

Alternatively, git-send-email will send a set of emails for you, but requires some configuration. Historically, git-send-email hasn't set proper email headers resulting in some mail servers rejecting the generated messages for various reasons.

3.5. What format should I use for the git pull request?

[rmk]
git pull requests should be formatted so that they contain the following information:

The commit ID which the patches are based upon
The URL and branch name (on the same line) to be pulled
The diffstat for the changes being submitted
A list of the changes contained within the pull

The 'git-request-pull' command generates a suitable pull message.

3.6. I have one patch. How should I submit it?

[rmk]
git can be problematical for submission of one patch - if the parent git commit is mischosen, or the pull request is of the wrong format, it takes time to resolve. A patch sent to the patch system is much more flexible - it doesn't have those issues. So, in some ways, having the patch outside of 'git' makes it quicker and easier to merge into git.

People listed above:

ds - Deepak Saxena
eb - Eric Benard
erikm - Erik Mouw
np - Nicolas Pitre
rfs - Ralph Siemsen
rmk - Russell King
rwhitby - Rod Whitby

Special thanks to Eric Bénard for volunteering to look after this FAQ.