[Dev] Re: embarrassing patch question
John F Davis
dev@trilug.org
Thu, 28 Feb 2002 15:56:31 -0500
Hello
Thanks for pointing out your man page opinions, but I read it a long time
ago and many times since.
Maybe I need to be clear on my question so I won't be given simple answers
like read the man page.
I'm asking because the man page is unclear as to how to use diff and patch
together to get
a good patch file and patch application method.
So one more time with feeling (and more details). I've got two directory
kernels in the current directory:
linux-2.4.2_hhl20 (The modified kernel directory) and
linux-2.4.2_hhl20.orig (The original kernel).
So, from the directory I issue the following command:
diff -Naur linxu-2.4.2_hhl20.orig linux-2.4.2_hhl20 > k2.patch
When the command finishes, I do the following commands to test the patch:
cp -rf linux-2.4.2_hhl20.orig linux-kernel-to-be-patched
(I'm giving it this weird name so I can be assured the name of the
directory doesn't matter.)
Then without changing the directory, I attempt to patch the directory like
so:
patch -d linux-to-be-patched -p1 < k2.patch
(Notice I'm doing the -p1 to ignore the leading dir name, and I'm using
the -d linux-to-be-patched option so it will use my new funky kernel
directory
name.)
The patch command runs like gangbusters, until the following occurs:
<leading output is snipped.>
patching file include/asm/posix_types.h
patching file include/asm/siginfo.h
patching file include/asm/unistd.h
patching file include/asm-ppc/posix_types.h
Reversed (or previously applied ) patch detected! Assume -R? [n]
And here is why that error occurs:
include/asm is a link to include/asm-ppc. So, when it applies
the patch to include/asm/posix_types.h it will fail when it does
include/asm-ppc/posix_types.h since its already been patched.
For what it matters, all three directories have the include/asm link to
include/asm-ppc
as they should.
Lastly, I wasn't trying to make a career out of diff/patch, I just wanted
to know how to do it
quickly and correctly. I figured it was a 5 minute problem and it only
needed a 5 minute question
of someone who knew how to do it to answer.
J (what the hell is a man page?) D
Ed Hill <ed@eh3.com>@trilug.org on 02/28/2002 03:09:08 PM
Please respond to dev@trilug.org
Sent by: dev-admin@trilug.org
To: dev@trilug.org
cc:
Subject: Re: [Dev] Re: embarrassing patch question
Hi John,
I think the patch man page does a fairly good job of explaining it:
-pnum or --strip=num
Strip the smallest prefix containing num leading
slashes from each file name found in the patch file. A
sequence of one or more adjacent slashes is counted as
a single slash. This controls how file names found in
the patch file are treated, in case you keep your files
in a different directory than the person who sent out
the patch. For example, supposing the file name in the
patch file was
/u/howard/src/blurfl/blurfl.c
setting -p0 gives the entire file name unmodified, -p1
gives
u/howard/src/blurfl/blurfl.c
without the leading slash, -p4 gives
blurfl/blurfl.c
and not specifying -p at all just gives you blurfl.c.
Whatever you end up with is looked for either in the
current directory, or the directory specified by the -d
option.
So, you specify a level in the -p option that is sufficient to strip
away enough leading path entries so that it matches the path that fits
with your current directory setup. Thus, you never *have* to have a
single file in the same directory name it was in when you created the
patch (if that patch is for a single file)! Extending the above
example, the pattern is:
-p0 ==> /u/howard/src/blurfl/blurfl.c
-p1 ==> u/howard/src/blurfl/blurfl.c
-p2 ==> howard/src/blurfl/blurfl.c
-p3 ==> src/blurfl/blurfl.c
-p4 ==> blurfl/blurfl.c
-p5 ==> blurfl.c
For a patch containing a set of changes to multiple files in a directory
tree composed of multiple levels of directories, you will want to keep
the directory names the same up to the root of the tree. That is, the
patch should apply cleanly if all the subdirs have the same names but
the root directory of the tree can have a different name (which you can
strip away using the appropriate -p[n] setting).
hth,
Ed
On Thu, 2002-02-28 at 12:46, John F Davis wrote:
>
> Hello
>
> Thanks for the reply, but my original question still exists.
>
> Here is your answer:
>
> 2.1 Create the patch
> diff -u oldfile newfile > filename.patch
> or diff -urN /olddir /newdir > filename.patch
> or diff -urN -X excludes /olddir /newdir > filename.patch
>
> diff options:
> -u Output (default 2) lines of unified context.
> -r Recursively compare any subdirectories found.
> -N Treat absent files as empty.
> -X FILE Exclude files that match any pattern in FILE.
>
> Note: The -u options can be replaced with -c to create a context format
> diff
> file with a setting of two lines.
>
> 2.2 Apply the patch
> gzip -cd patch.gz | patch -p0
> or bzip2 -dc patch.bz2 | patch -p0
> or patch -p0 < filename.patch
>
> patch options:
> -p NUM Strip NUM leading components from file names.
>
>
> Here is my original question:
>
>
> > I think this is right. However, it would be great if someone would
clear
> > up the bit
> > about the target directory. Does the target directory have to called
the
> > same as the
> > "patched" or "orig" directory? Is there a way to specify or edit the
> patch
> > file so
> > that the end user can have their own custom directory name. For
> instance,
> > I used foo and foo.orig, but the end user might have directory names
> > like foo2 or foo.nfs, etc.
> >
>
> Put another way, in your step 2.2, what is the implied directory name
used
> for the patch? What is the p1 or p0 bit? Is that supposed to
> somehow strip the leading direcotry name or something? When you issue
that
> comand, are you in the parent directory which owns
> the kernel directory to be patched or are you in the directory to be
> patched?
>
> eg.:
>
> ls /usr/local/
> foo foo.orig
>
> cd foo
> patch -p0 < patchfile
>
> or
>
> ls /usr/local
> foo foo.orig patchfile
> patch -p0 < patchfile
>
>
>
> David Filion <filiond@videotron.ca> on 02/28/2002 11:39:42 AM
>
>
>
>
>
> To: John F Davis/Raleigh/IBM@ibmus
> cc: dev@trilug.org, kernelnewbies@nl.linux.org
> Subject: Re: embarrassing patch question
>
>
>
> John F Davis wrote:
> >
> > Hello
> >
> > Ok. I've been using patch and I've asked some questions about it on
IRC,
> > but I'm still not
> > sure about how to use it "CORRECTLY" or in the "PRESCRIBED-METHOD".
So,
> > please correct me.
> >
> > I want to create a patch file to give to my co-workers and instructions
> on
> > how to use it.
> >
> > With that said, here is the scenario:
> >
> > I have a original directory called foo.orig and a "up-to-date"
directory
> > called foo.latest.
> > So, I created a patch file using this syntax:
> > diff -Naur foo.orig foo.latest > foo.patch
> >
> > Then to apply the patch, I did:
> > patch -p1 foo < foo.patch
> > or
> > patch -p1 foo.orig < foo.patch
> > I can't remember which one I did.
> >
> > I think this is right. However, it would be great if someone would
clear
> > up the bit
> > about the target directory. Does the target directory have to called
the
> > same as the
> > "patched" or "orig" directory? Is there a way to specify or edit the
> patch
> > file so
> > that the end user can have their own custom directory name. For
> instance,
> > I used foo and foo.orig, but the end user might have directory names
> > like foo2 or foo.nfs, etc.
> >
> > JD
> >
> > --
> > Kernelnewbies: Help each other learn about the Linux kernel.
> > Archive: http://mail.nl.linux.org/kernelnewbies/
> > IRC Channel: irc.openprojects.net / #kernelnewbies
> > Web Page: http://www.kernelnewbies.org/
>
>
> I threw together something a while ago when I was in the same
> situation, I've included it below. Because we are using Solaris
> at work, it covers the older and newer versions of diff/patch.
>
> HTH
> David Filion
>
> --- Start of doc ---
>
> Creating a patch
>
> Below is a cheat sheet for creating and applying patch files. For more
> details about each command, check the commands manpage, info document or
> try "command --help".
>
>
> 1.0 Using older versions of diff/patch.
> The older diff/patch utilities don't understand the newer unified
> format. Thus, the older conext format must be used. Thus the -c
> option that appears.
>
> 1.1 Create the patch
> diff -c -b -r /original /new > filename.patch
>
> diff options:
> -c Produces a listing of differences with three lines of context.
> -b Ignores trailing blanks (spaces and tabs) and treats other strings
> of blanks as equivalent.
> -r Applies diff recursively to common subdirectories encountered.
>
> 1.2 Apply the patch
> patch -c -l -p1 -d /original < filename.patch
>
> patch options:
> -c Interpret the patch file as a context difference (the output of
> the command diff when the -c or -C options are specified).
> -l Cause any sequence of blank characters in the difference script
> to match any sequence of blank characters in the input file.
> Other characters will be matched exactly.
> -d Change the current directory to dir before processing.
> -p NUM Strip NUM leading components from file names.
>
> 2.0 Using newer/GNU diff and patch utilities.
> Unlike their older counterparts, the newer/GNU versions of diff
> understand the unified format of patch file. This is the format used
> by the Linux kernel developers.
>
> 2.1 Create the patch
> diff -u oldfile newfile > filename.patch
> or diff -urN /olddir /newdir > filename.patch
> or diff -urN -X excludes /olddir /newdir > filename.patch
>
> diff options:
> -u Output (default 2) lines of unified context.
> -r Recursively compare any subdirectories found.
> -N Treat absent files as empty.
> -X FILE Exclude files that match any pattern in FILE.
>
> Note: The -u options can be replaced with -c to create a context format
> diff
> file with a setting of two lines.
>
> 2.2 Apply the patch
> gzip -cd patch.gz | patch -p0
> or bzip2 -dc patch.bz2 | patch -p0
> or patch -p0 < filename.patch
>
> patch options:
> -p NUM Strip NUM leading components from file names.
>
> 3.0 Creating a patch using CVS.
> A patch is created in CVS using the rdiff cvs command.
>
> 3.1 Create the patch.
>
> o Create a context (default) format diff of release tags rel1 and
> rel2 for module.
> cvs rdiff -r rel1 -r rel2 module
>
> o Create a unidiff format diff of release tags rel1 and rel2
> for module.
> cvs rdiff -u -r rel1 -r rel2 module
>
> cvs rdiff options:
> -r tag Use revision <tag>
> -u Use the unidiff format.
> -c Use the context format (default).
>
>
> Copyright (c) David Filion (filiond@videotron.ca)
> Created 2001/09/12.
>
>
> _______________________________________________
> Dev mailing list
> Dev@trilug.org
> http://www.trilug.org/mailman/listinfo/dev
>
--
Edward H. Hill III, PhD
Post-Doctoral Researcher | Email: ed@eh3.com, ehill@mines.edu
Division of ESE | URL: http://www.eh3.com
Colorado School of Mines | Phone: 303-273-3483
Golden, CO 80401 | Fax: 303-273-3311
Key fingerprint = 5BDE 4DA1 66BE 4F7B BC17 3A0C 932B 7266 1E76 F123
_______________________________________________
Dev mailing list
Dev@trilug.org
http://www.trilug.org/mailman/listinfo/dev