Applying kernel patches questions
 

I'm pretty new at compiling kernels, and am a bit confused about a basic point. Could somebody please enlighten me as to how to apply the 2.5.69-mm7.bz2 kernel "patch"?

I undersand and have applied other patches such as ones with a "patch" in the name, like ck_2.4.20.patch.bz2, and patch-2.5.69-ac1.bz2. Since these mm patches don't have the "patch" in the name, do I use the same command format for applying them as for other patches- and do I have to apply all the others in succession, like mm2, mm3, mm4 etc., up to the last current mm7?

What would help the most would be if I could see the exact correct command used when applying the mm kernel patches.

Many thanks,
wrc1944

If you're unsure if it is a patch, just bunzip2 2.5.69-mm7.bz2 and "file or less" the contents. If it seems to be a patch just "bzcat 2.5.69-mm7.bz2 | patch <args>".

I extracted it, and it gave me a 7.5MB text file, and I have included the top part. How can I tell if it actually is a kernel patch. I assume it is, because Andrew Morton is one of those who issues kernel patches, and I got it at the kernel patch page. Why wouldn't they just say it's a patch, in other words, make it perfectly clear as most other kernel patches are? It's very confusing for kernel newbies if some patches have "patch" in the name, and some don't.

Am I correct in assuming that 2.5.69-mm7.bz2 is indeed a patch, even though there is no "patch" in it's name, and I can treat it as such, and apply it as I have done others?

wrc1944


--- linux-2.5.69/arch/alpha/kernel/core_marvel.c 2003-05-04 21:18:12.000000000 -0700
+++ 25/arch/alpha/kernel/core_marvel.c 2003-05-17 14:49:04.000000000 -0700
@@ -780,7 +780,7 @@ __marvel_rtc_io(int write, u8 b, unsigne
rtc_access.function = 0x49; /* GET_TOY */
if (write) rtc_access.function = 0x48; /* PUT_TOY */

-#if CONFIG_SMP
+#ifdef CONFIG_SMP
if (smp_processor_id() != boot_cpuid)
smp_call_function_on_cpu(__marvel_access_rtc,
&rtc_access,
--- linux-2.5.69/arch/alpha/kernel/entry.S 2003-03-04 20:02:35.000000000 -0800
+++ 25/arch/alpha/kernel/entry.S 2003-05-17 14:49:04.000000000 -0700
@@ -849,7 +849,7 @@ sys_getxpid:
about this loop. */

I extracted it, and it gave me a 7.5MB text file, and I have included the top part.
That's usefull info, thnx for posting.

How can I tell if it actually is a kernel patch.
Making a patch is done comparing (diff) two files, like in "diff -urN <file_0> <file_1>" The first line corresponds with the old file (version), the second one with the new one. Reading the first two line you notice the triple minus and plusses, a filename and a date on each line. In the content you'll notice plusses and minusses and filenames for each file which "diff" detected changed contents. That's the only way to determine it's a patch AFAIK, cuz using "file" on it will only say something about "std ASCII C text".

I assume it is, because Andrew Morton is one of those who issues kernel patches, and I got it at the kernel patch page.
Even tho the location and the name fit the bill, the best way (strict) to determine if a patch is issued by someone would be to have verification, like it being GPG-signed (like the kernel sources, or the GRSecurity kernel patches have). Else you'll just have to trust 'em. Because noone really enforces GPG signing it's unfortunately easy for ppl with malicious intent to inject their code into sources (see Bitchx, TCP Wrappers, OpenSSH, Aide etc etc). Of course it will be uncovered, but only by the first (few) who actually read code, and notice differences in behaviour, and notice awkward in/outbound traffic, and post it.

Anyway, nuff ranting, this seems to be a patch, so you know what to do...

Went ahead and applied it to 2.5.69, and compiled and installed with no warnings or errors! This is the first time I've ever gotten through a 2.5.xx kernel without many warnings. I'm trying this on Mandrake 9.1. I'm currently in 2.5.69-mm7 asI type.

Everthing is OK, except I still have the problem I had with 2.5.67, and 2.5.68, and about 25 tries with various configs on 2.5.69.

Briefly, it's narrowed down to this:
If I enable serial drivers in xconfig, it compiles, and I can get on the internet fine- works great. But when I use other apps, it takes 5 minutes to save or copy/paste any documents, or open various programs. This occurs with kde, or any other GUI, and sometimes in consoles. I also cannot open /home- if it does open, it can take up to 8-9 minutes. he strange thing is, konqueror as a browser seems to work great, but any "save as" operation will proceed for at least 3-4 minutes- everything else is lightning fast. Weird!

If I compile 2.5.69 without enabling serial drivers, everything seems to work correctly and very fast, but then I can't use my serial modem- it's as if it doesn't exist.

Any ideas out there? I've asked about this on other forums, and so far nobody has figured it out. When this mm7 version of 2.5.69 compiled with no warnings whatsoever, I thought I had finally gotten it to work, but alas, same deal as before.

wrc1944