cvs commit: hints nptl.txt

timothy at linuxfromscratch.org timothy at linuxfromscratch.org
Mon May 5 20:51:12 PDT 2003


timothy     03/05/05 23:51:12

  Modified:    .        nptl.txt
  Log:
  Added fix to allow building against old hosts. Added patch to fix detection of eh_frame using coreutils. Added note about rml's procps.
  
  Revision  Changes    Path
  1.9       +31 -0     hints/nptl.txt
  
  Index: nptl.txt
  ===================================================================
  RCS file: /home/cvsroot/hints/nptl.txt,v
  retrieving revision 1.8
  retrieving revision 1.9
  diff -u -r1.8 -r1.9
  --- nptl.txt	26 Apr 2003 16:21:22 -0000	1.8
  +++ nptl.txt	6 May 2003 03:51:12 -0000	1.9
  @@ -29,6 +29,9 @@
     * 0.6 - 20030423 - Moved to GCC 3.3.
     * 0.7 - 20030424 - w00t. NVIDIA_GLX works now.
     * 0.8 - 20030426 - Devfs notes. Patch to make 2.4.20 work w/ NPTL (Thanks YuX)
  +  * 0.9 - 20030505 - Added fix to allow building against old hosts. Added patch
  +		     to fix detection of eh_frame using coreutils. Added note
  +		     about rml's procps.
   
   
   Introduction
  @@ -156,12 +159,21 @@
   Ch.5 -- GCC
   -----------
   
  +Apply the gcc-3.3-coreutils patch. This fixes some deprecrated syntax to the
  +head and tail commands, one of which will cause the configure script to fail
  +in detecting eh_frame optimization support, therefore compromising the build.
  +
   Some tests will fail. It is recommended but not necessary that you examine the
   testsuite logs to see why the tests failed, in addition to examining each test
   case to decide if it really matters to you. Last time I checked, I had some 9
   different tests fail, and of those 9, only 2 of them were actual failures, the
   rest were outdated or broken test cases.
   
  +During any of the Chapter 5 GCC builds, you may run into an error somewhat like
  +the following: "storage class specified for parameter `type name'" or something
  +of that sort. The cause of this is deprecated use of the __thread keyword,
  +and can be fix by sed'ing all of the __thread occurances to __thread_old.
  +
   
   Ch.6 -- Kernel headers
   ----------------------
  @@ -193,6 +205,25 @@
   
   If you are using a 2.5 kernel, it's safe to say that you have no reason at all
   to install this package. Just install module-init-tools in its place.
  +
  +
  +Ch.6 -- Procps
  +--------------
  +
  +Since you're likely using a development kernel, it's safe to say that you'd want
  +a good stable procps package to read it's (sometimes incompatible) /proc fs. The
  +version of procps that you probably want to use is located here:
  +
  +	http://tech9.net/rml/procps
  +
  +While it could be said that this version of procps is inferious because it uses
  +the "old" codebase, you must look at the argument from the other side of the
  +issue, and look closer at the other version (procps 3.x.xx). The older code has
  +been proven time and time again, as opposed to the mostly rewritten version,
  +which has been bugged beyond belief since its inception. For example, there was
  +a bug that lived for who know's how long that prevented procps from showing
  +__ANY__ pid's starting with the number 6. Just imagine what hasn't been found
  +yet.
   
   
   Ch.6 -- Sysklogd
  
  
  
-- 
Unsubscribe: send email to listar at linuxfromscratch.org
and put 'unsubscribe hints' in the subject header of the message



More information about the hints mailing list