Re: usbkey, GCC version for kernel compilation
On Tue, May 09, 2006 at 02:31:23AM +0300, Chaim Keren Tzion wrote: ... My recommendation: Exchange the usbkey for a new one. There is a 255 to 1 chance that the replacement, even the same brand, will be okay. You can also just recompile your kernel and remove the cumana support but one day you may want to use the usbkey on another machine that you will discover has this obscure partition type enabled in its kernel and you will be usb-diskless, so to speak. That will not be a happy day. -- Chaim Keren Tzion [EMAIL PROTECTED] +972-54-811-9234 Thanks Chaim. Icannot follow your advice: I got this key as a birthday, a couple of months ago. But the danger that I'll try to use the key on another linux system presenting the same problem is very low in my case... I shall recompile the kernel, instead. Now, in the documentation coming with the kernel source they recommend to use GCC vesrion 2.95.3. That is teribly old, and probably hard to get. I guesss the documentation is not updated. I recall that there was a time, 2 or 3 years ago when they recomended to use the old 2.95.x version of gcc to compile the kernel, while using the newer version for everything else. This is, probably, long time passe. But, to be on the safe side, what is the present recommendation? Cheers, Avraham = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: usbkey, GCC version for kernel compilation
On 5/15/06, Avraham Rosenberg [EMAIL PROTECTED] wrote: On Tue, May 09, 2006 at 02:31:23AM +0300, Chaim Keren Tzion wrote: ... My recommendation: Exchange the usbkey for a new one. There is a 255 to 1 chance that the replacement, even the same brand, will be okay. You can also just recompile your kernel and remove the cumana support but one day you may want to use the usbkey on another machine that you will discover has this obscure partition type enabled in its kernel and you will be usb-diskless, so to speak. That will not be a happy day. -- Chaim Keren Tzion [EMAIL PROTECTED] +972-54-811-9234 Thanks Chaim. Icannot follow your advice: I got this key as a birthday, a couple of months ago. But the danger that I'll try to use the key on another linux system presenting the same problem is very low in my case... I shall recompile the kernel, instead. Now, in the documentation coming with the kernel source they recommend to use GCC vesrion 2.95.3. That is teribly old, and probably hard to get. I guesss the documentation is not updated. I recall that there was a time, 2 or 3 years ago when they recomended to use the old 2.95.x version of gcc to compile the kernel, while using the newer version for everything else. This is, probably, long time passe. But, to be on the safe side, what is the present recommendation? Indeed README still talks about 2.95.3, which was latest stable many years ago :) When you're not sure which version to use the best practice is to use default gcc which came with your distribution: it should be compatible with your kernel. My 'gcc -v' reports version 3.4.5 and produced working 2.6 kernel images. Another advice is to reuse existing kernel configuration from /proc/config.gz, just change relevant kernel option and use 'make oldconfig' build procedure. Cheers, Avraham = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED] -- Alexander Indenbaum To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: usbkey, GCC version for kernel compilation
On Mon, May 15, 2006 at 11:28:24PM +0300, Alexander Indenbaum wrote: ... Another advice is to reuse existing kernel configuration from /proc/config.gz, just change relevant kernel option and use 'make oldconfig' build procedure. -- Alexander Indenbaum Good morning, Alexander and thanks. There is no /proc/config.gz in my system, maybe because I did not compile the present kernel - I used the kernel from the installing CD. I found the present configuration in /boot/config-2.6.8-2-386, instead. The other advice is better than Muli's suggestion, found in an old mail (he suggested making first the 'make oldconfig' and then editing the resulted .config file), because it takes advantage of the control mechanism of the config command. But I have, first, to do some reading about the various options and their influence with kernel 2.6.8. It is a long time since I last compiled my kernel, and then there are so many options added. Cheers, Avraham = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Kernel compilation oddities
On Sunday 05 March 2006 00:12, Muli Ben-Yehuda wrote: every time I compile the kernel it takes amazingly long, and goes through a full compilation. Even running 'make' twice in a row makes it compile everything again. make version? assuming it's 3.81rc1, it's a known make bug - see http://marc.theaimsgroup.com/?l=linux-kernelm=114150857516839w=2 Yup. # dpkg -l | grep make ii make 3.80+3.81.rc1-1 The GNU version of the make utility. I guess I'll have to downgrade (until kbuild changes). Cheers, Muli Thanks :-) - Aviram = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Kernel compilation oddities
On Sat, Mar 04, 2006 at 09:02:30AM +0200, Aviram Jenik wrote: Now, I'm on another laptop, and trying to do the same. However, every time I compile the kernel it takes amazingly long, and goes through a full compilation. Even running 'make' twice in a row makes it compile everything again. This happened to me with 2.6.13.1 and 2.6.15 and 2.6.15.4. What should I check? I'm running Debian SID, BTW. make version? assuming it's 3.81rc1, it's a known make bug - see http://marc.theaimsgroup.com/?l=linux-kernelm=114150857516839w=2 Cheers, Muli -- Muli Ben-Yehuda http://www.mulix.org | http://mulix.livejournal.com/ = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Kernel compilation oddities
On Sunday March 5 2006 00:12, Muli Ben-Yehuda wrote: On Sat, Mar 04, 2006 at 09:02:30AM +0200, Aviram Jenik wrote: Now, I'm on another laptop, and trying to do the same. However, every time I compile the kernel it takes amazingly long, and goes through a full compilation. Even running 'make' twice in a row makes it compile everything again. This happened to me with 2.6.13.1 and 2.6.15 and 2.6.15.4. What should I check? I'm running Debian SID, BTW. make version? assuming it's 3.81rc1, it's a known make bug - see http://marc.theaimsgroup.com/?l=linux-kernelm=114150857516839w=2 Now that's evil to break it like that. Am I right to summarize it to specific incompatibility of version 3.81rc1 and kbuild script and that it is the script that is about to be changed, not the make? I am concerned about the possibility to eventually pulling a brand new make into the system while still having an old kernel source tree. Looks like Debian SID users have already stepped on that rake. -- Sincerely Yours, Michael Vasiliev In order to live free and happily, you must sacrifice boredom. It is not always an easy sacrifice. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Kernel compilation oddities
On Sun, Mar 05, 2006 at 03:17:26AM +0200, Michael Vasiliev wrote: Now that's evil to break it like that. Am I right to summarize it to specific incompatibility of version 3.81rc1 and kbuild script and that it is the script that is about to be changed, not the make? Let's put it this way: make has changed its behaviour from an unspecified one to an unspecified but arguably more correct one. Kbuild was relying on the old behaviour and will be fixed to work with the new one. I am concerned about the possibility to eventually pulling a brand new make into the system while still having an old kernel source tree. Looks like Debian SID users have already stepped on that rake. .. and the universe blow up? not quite, you just suffer a few full recompilations. I'm sure someone will come up with the appropriate fix for every kernel version. Cheers, Muli -- Muli Ben-Yehuda http://www.mulix.org | http://mulix.livejournal.com/ = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Kernel compilation oddities
Hi, On my previous laptop, I would compile the kernel once, and then if I need to compile the same kernel again (lets say I only changed something from being compiled-in to being a module) I would run make, and watch it skip the already-compiled parts quite quickly. That way, recompiling an existing kernel would be much faster comparing to compiling a 'fresh' kernel, except for cases where a change affects most of the compilation (e.g. changing a setting that affects all modules). Now, I'm on another laptop, and trying to do the same. However, every time I compile the kernel it takes amazingly long, and goes through a full compilation. Even running 'make' twice in a row makes it compile everything again. This happened to me with 2.6.13.1 and 2.6.15 and 2.6.15.4. What should I check? I'm running Debian SID, BTW. Thanks, Aviram = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
kernel compilation
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Hi, I'm trying to compile kernel 2.6.8-1 with debian kernel-package. I think i'm missing libraries. Can someone please point me of the libs or paths i need to add. tnx kfir -BEGIN PGP SIGNATURE- Version: GnuPG v1.2.4 (GNU/Linux) iD8DBQFBlhRMe7jKk87FUO8RAmY8AJ9tx4jVbkBtoz/hmIb39gkYrsyhmgCfeJ0Z rUbaO77h+6GSnEzIJGxXaG8= =j6dd -END PGP SIGNATURE- /usr/bin/make\ ARCH=i386 menuconfig make[1]: Entering directory `/usr/src/kernel-source-2.6.8' HOSTCC scripts/basic/fixdep scripts/basic/fixdep.c:97:23: sys/types.h: No such file or directory scripts/basic/fixdep.c:98:22: sys/stat.h: No such file or directory scripts/basic/fixdep.c:99:22: sys/mman.h: No such file or directory scripts/basic/fixdep.c:100:20: unistd.h: No such file or directory scripts/basic/fixdep.c:101:19: fcntl.h: No such file or directory scripts/basic/fixdep.c:102:20: string.h: No such file or directory scripts/basic/fixdep.c:103:20: stdlib.h: No such file or directory scripts/basic/fixdep.c:104:19: stdio.h: No such file or directory In file included from /usr/lib/gcc-lib/i486-linux/3.3.4/include/syslimits.h:7, from /usr/lib/gcc-lib/i486-linux/3.3.4/include/limits.h:11, from scripts/basic/fixdep.c:105: /usr/lib/gcc-lib/i486-linux/3.3.4/include/limits.h:122:75: limits.h: No such file or directory scripts/basic/fixdep.c:106:19: ctype.h: No such file or directory scripts/basic/fixdep.c:107:24: netinet/in.h: No such file or directory scripts/basic/fixdep.c: In function `usage': scripts/basic/fixdep.c:121: warning: implicit declaration of function `fprintf' scripts/basic/fixdep.c:121: error: `stderr' undeclared (first use in this function) scripts/basic/fixdep.c:121: error: (Each undeclared identifier is reported only once scripts/basic/fixdep.c:121: error: for each function it appears in.) scripts/basic/fixdep.c:122: warning: implicit declaration of function `exit' scripts/basic/fixdep.c: In function `print_cmdline': scripts/basic/fixdep.c:127: warning: implicit declaration of function `printf' scripts/basic/fixdep.c: At top level: scripts/basic/fixdep.c:130: error: `NULL' undeclared here (not in a function) scripts/basic/fixdep.c: In function `grow_config': scripts/basic/fixdep.c:143: warning: implicit declaration of function `realloc' scripts/basic/fixdep.c:143: warning: assignment makes pointer from integer without a cast scripts/basic/fixdep.c:144: error: `NULL' undeclared (first use in this function) scripts/basic/fixdep.c:145: warning: implicit declaration of function `perror' scripts/basic/fixdep.c: In function `is_defined_config': scripts/basic/fixdep.c:161: warning: implicit declaration of function `memcmp' scripts/basic/fixdep.c: In function `define_config': scripts/basic/fixdep.c:174: warning: implicit declaration of function `memcpy' scripts/basic/fixdep.c: In function `use_config': scripts/basic/fixdep.c:193: error: `PATH_MAX' undeclared (first use in this function) scripts/basic/fixdep.c:207: warning: implicit declaration of function `tolower' scripts/basic/fixdep.c:193: warning: unused variable `s' scripts/basic/fixdep.c: At top level: scripts/basic/fixdep.c:212: error: parse error before size_t scripts/basic/fixdep.c:213: warning: function declaration isn't a prototype scripts/basic/fixdep.c: In function `parse_config_file': scripts/basic/fixdep.c:214: error: `map' undeclared (first use in this function) scripts/basic/fixdep.c:214: error: `len' undeclared (first use in this function) scripts/basic/fixdep.c:220: warning: implicit declaration of function `ntohl' scripts/basic/fixdep.c:231: warning: implicit declaration of function `isalnum' scripts/basic/fixdep.c: In function `strrcmp': scripts/basic/fixdep.c:244: warning: implicit declaration of function `strlen' scripts/basic/fixdep.c: In function `do_config_file': scripts/basic/fixdep.c:255: error: storage size of `st' isn't known scripts/basic/fixdep.c:259: warning: implicit declaration of function `open' scripts/basic/fixdep.c:259: error: `O_RDONLY' undeclared (first use in this function) scripts/basic/fixdep.c:261: error: `stderr' undeclared (first use in this function) scripts/basic/fixdep.c:265: warning: implicit declaration of function `fstat' scripts/basic/fixdep.c:267: warning: implicit declaration of function `close' scripts/basic/fixdep.c:270: warning: implicit declaration of function `mmap' scripts/basic/fixdep.c:270: error: `NULL' undeclared (first use in this function) scripts/basic/fixdep.c:270: error: `PROT_READ' undeclared (first use in this function) scripts/basic/fixdep.c:270: error: `MAP_PRIVATE' undeclared (first use in this function) scripts/basic/fixdep.c:270: warning: assignment makes pointer from integer without a cast scripts/basic/fixdep.c:279: warning: implicit declaration of function `munmap' scripts/basic/fixdep.c:255: warning: unused variable `st' scripts/basic/fixdep.c: At top level:
Re: kernel compilation
On Sat, Nov 13, 2004 at 04:03:47PM +0200, Kfir Lavi wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Hi, I'm trying to compile kernel 2.6.8-1 with debian kernel-package. I think i'm missing libraries. Can someone please point me of the libs or paths i need to add. Where did you obtain the source from? You might want to use the kernel-package. Although probably not related to the problem you mentioned, the whole process might be easier with it. tnx kfir -BEGIN PGP SIGNATURE- Version: GnuPG v1.2.4 (GNU/Linux) iD8DBQFBlhRMe7jKk87FUO8RAmY8AJ9tx4jVbkBtoz/hmIb39gkYrsyhmgCfeJ0Z rUbaO77h+6GSnEzIJGxXaG8= =j6dd -END PGP SIGNATURE- /usr/bin/make\ ARCH=i386 menuconfig make[1]: Entering directory `/usr/src/kernel-source-2.6.8' HOSTCC scripts/basic/fixdep scripts/basic/fixdep.c:97:23: sys/types.h: No such file or directory scripts/basic/fixdep.c:98:22: sys/stat.h: No such file or directory scripts/basic/fixdep.c:99:22: sys/mman.h: No such file or directory scripts/basic/fixdep.c:100:20: unistd.h: No such file or directory scripts/basic/fixdep.c:101:19: fcntl.h: No such file or directory scripts/basic/fixdep.c:102:20: string.h: No such file or directory scripts/basic/fixdep.c:103:20: stdlib.h: No such file or directory scripts/basic/fixdep.c:104:19: stdio.h: No such file or directory In file included from /usr/lib/gcc-lib/i486-linux/3.3.4/include/syslimits.h:7, from /usr/lib/gcc-lib/i486-linux/3.3.4/include/limits.h:11, from scripts/basic/fixdep.c:105: /usr/lib/gcc-lib/i486-linux/3.3.4/include/limits.h:122:75: limits.h: No such file or directory scripts/basic/fixdep.c:106:19: ctype.h: No such file or directory scripts/basic/fixdep.c:107:24: netinet/in.h: No such file or directory scripts/basic/fixdep.c: In function `usage': scripts/basic/fixdep.c:121: warning: implicit declaration of function `fprintf' scripts/basic/fixdep.c:121: error: `stderr' undeclared (first use in this function) scripts/basic/fixdep.c:121: error: (Each undeclared identifier is reported only once scripts/basic/fixdep.c:121: error: for each function it appears in.) scripts/basic/fixdep.c:122: warning: implicit declaration of function `exit' scripts/basic/fixdep.c: In function `print_cmdline': scripts/basic/fixdep.c:127: warning: implicit declaration of function `printf' scripts/basic/fixdep.c: At top level: scripts/basic/fixdep.c:130: error: `NULL' undeclared here (not in a function) scripts/basic/fixdep.c: In function `grow_config': scripts/basic/fixdep.c:143: warning: implicit declaration of function `realloc' scripts/basic/fixdep.c:143: warning: assignment makes pointer from integer without a cast scripts/basic/fixdep.c:144: error: `NULL' undeclared (first use in this function) scripts/basic/fixdep.c:145: warning: implicit declaration of function `perror' scripts/basic/fixdep.c: In function `is_defined_config': scripts/basic/fixdep.c:161: warning: implicit declaration of function `memcmp' scripts/basic/fixdep.c: In function `define_config': scripts/basic/fixdep.c:174: warning: implicit declaration of function `memcpy' scripts/basic/fixdep.c: In function `use_config': scripts/basic/fixdep.c:193: error: `PATH_MAX' undeclared (first use in this function) scripts/basic/fixdep.c:207: warning: implicit declaration of function `tolower' scripts/basic/fixdep.c:193: warning: unused variable `s' scripts/basic/fixdep.c: At top level: scripts/basic/fixdep.c:212: error: parse error before size_t scripts/basic/fixdep.c:213: warning: function declaration isn't a prototype scripts/basic/fixdep.c: In function `parse_config_file': scripts/basic/fixdep.c:214: error: `map' undeclared (first use in this function) scripts/basic/fixdep.c:214: error: `len' undeclared (first use in this function) scripts/basic/fixdep.c:220: warning: implicit declaration of function `ntohl' scripts/basic/fixdep.c:231: warning: implicit declaration of function `isalnum' scripts/basic/fixdep.c: In function `strrcmp': scripts/basic/fixdep.c:244: warning: implicit declaration of function `strlen' scripts/basic/fixdep.c: In function `do_config_file': scripts/basic/fixdep.c:255: error: storage size of `st' isn't known scripts/basic/fixdep.c:259: warning: implicit declaration of function `open' scripts/basic/fixdep.c:259: error: `O_RDONLY' undeclared (first use in this function) scripts/basic/fixdep.c:261: error: `stderr' undeclared (first use in this function) scripts/basic/fixdep.c:265: warning: implicit declaration of function `fstat' scripts/basic/fixdep.c:267: warning: implicit declaration of function `close' scripts/basic/fixdep.c:270: warning: implicit declaration of function `mmap' scripts/basic/fixdep.c:270: error: `NULL' undeclared (first use in this function) scripts/basic/fixdep.c:270: error: `PROT_READ' undeclared (first use in this function)
Re: kernel compilation
On Saturday 13 November 2004 16:03, Kfir Lavi wrote: Hi, I'm trying to compile kernel 2.6.8-1 with debian kernel-package. I think i'm missing libraries. Can someone please point me of the libs or paths i need to add. Hi Kfir! sys/types.h, sys/stat.h, unistd.h, etc. are all standard UNIX C header files. If gcc cannot find them, then you probably need to install the glibc development package: http://packages.debian.org/stable/devel/libc6-dev Regards, Shlomi Fish - Shlomi Fish [EMAIL PROTECTED] Homepage:http://www.shlomifish.org/ Knuth is not God! It took him two days to build the Roman Empire. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
Kfir Lavi wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Hi, I'm trying to compile kernel 2.6.8-1 with debian kernel-package. I think i'm missing libraries. Can someone please point me of the libs or paths i need to add. tnx kfir -BEGIN PGP SIGNATURE- Version: GnuPG v1.2.4 (GNU/Linux) iD8DBQFBlhRMe7jKk87FUO8RAmY8AJ9tx4jVbkBtoz/hmIb39gkYrsyhmgCfeJ0Z rUbaO77h+6GSnEzIJGxXaG8= =j6dd -END PGP SIGNATURE- Hi Kfir. You missed libc6-dev. Better use kernel sources with debian patches: apt-get install libc6-dev kernel-source-2.6.8 Yury. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
Hi On Sat, Nov 13, 2004 at 04:03:47PM +0200, Kfir Lavi wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Hi, I'm trying to compile kernel 2.6.8-1 with debian kernel-package. I think i'm missing libraries. Can someone please point me of the libs or paths i need to add. /usr/bin/make\ ARCH=i386 menuconfig make[1]: Entering directory `/usr/src/kernel-source-2.6.8' HOSTCC scripts/basic/fixdep scripts/basic/fixdep.c:97:23: sys/types.h: No such file or directory scripts/basic/fixdep.c:98:22: sys/stat.h: No such file or directory scripts/basic/fixdep.c:99:22: sys/mman.h: No such file or directory scripts/basic/fixdep.c:100:20: unistd.h: No such file or directory scripts/basic/fixdep.c:101:19: fcntl.h: No such file or directory scripts/basic/fixdep.c:102:20: string.h: No such file or directory scripts/basic/fixdep.c:103:20: stdlib.h: No such file or directory scripts/basic/fixdep.c:104:19: stdio.h: No such file or directory In file included from /usr/lib/gcc-lib/i486-linux/3.3.4/include/syslimits.h:7, from /usr/lib/gcc-lib/i486-linux/3.3.4/include/limits.h:11, from scripts/basic/fixdep.c:105: /usr/lib/gcc-lib/i486-linux/3.3.4/include/limits.h:122:75: limits.h: No such file or directory http://www.debian.org/distrib/packages#search_contents to the rescue! http://packages.debian.org/cgi-bin/search_contents.pl?word=sys%2Ftypes.hsearchmode=searchfilescase=sensitiveversion=unstablearch=i386 -- Tzafrir Cohen +---+ http://www.technion.ac.il/~tzafrir/ |vim is a mutt's best friend| mailto:[EMAIL PROTECTED] +---+ = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
Tzafrir Cohen wrote: http://www.debian.org/distrib/packages#search_contents to the rescue! Or better yet (IMHO) - apt-file search. http://packages.debian.org/cgi-bin/search_contents.pl?word=sys%2Ftypes.hsearchmode=searchfilescase=sensitiveversion=unstablearch=i386 apt-file search sys/types.h Cheers, --Amos = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
[EMAIL PROTECTED] wrote: Tzafrir Cohen wrote: http://www.debian.org/distrib/packages#search_contents to the rescue! Or better yet (IMHO) - apt-file search. http://packages.debian.org/cgi-bin/search_contents.pl?word=sys%2Ftypes.hsearchmode=searchfilescase=sensitiveversion=unstablearch=i386 apt-file search sys/types.h Cheers, Even better: apt-get build-dep kernel-image-2.6.8-1-686 It installs all the packages that are needed if you want to build the package you specify. Shachar -- Shachar Shemesh Lingnu Open Source Consulting ltd. http://www.lingnu.com/ = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Solved! I needed to install the packages: libc6-dev kernel-heders libncurses5-dev Thanks to all the people that helped me. kfir On Saturday 13 November 2004 16:03, Kfir Lavi wrote: Hi, I'm trying to compile kernel 2.6.8-1 with debian kernel-package. I think i'm missing libraries. Can someone please point me of the libs or paths i need to add. tnx kfir -BEGIN PGP SIGNATURE- Version: GnuPG v1.2.4 (GNU/Linux) iD8DBQFBlpyKe7jKk87FUO8RAqVbAJ4+OqJXUuyriVC/KBgYza5nw9iRlgCgkozy fmdO9T08vh7MnXGsELH9jg0= =Ua5B -END PGP SIGNATURE- To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
Thank you all very much for your help. With the knowledge you gave me I found out: * The AES encryption is already compiled by RH distro * It is compiled as a module * When looking at /proc/crypto/cipher/ and using one of the encryptions found there the 'losetup' command works (the one from util-linux-2.12, not the distro one) In the next couple of days I need to make a change in the kernel and then compile and install it. I hope I will manage to do that with my knowledge (it seems simple when reading about it), but I'm sure I will get your help if I need it, right? Well, thanks again, David. From: Muli Ben-Yehuda [EMAIL PROTECTED] To: David Sapir [EMAIL PROTECTED] CC: [EMAIL PROTECTED], [EMAIL PROTECTED] Subject: Re: kernel compilation Date: Tue, 9 Mar 2004 17:33:15 +0200 On Tue, Mar 09, 2004 at 04:34:53PM +0200, David Sapir wrote: How do I know which of the config files were used to compile my kernel (the original distro kernel)? Usually, there's correspondence between the kernel version of the running kernel (try 'uname -a' to see it) and the config file's name. For example, on the machine that has kernel '2.4.22-1.2115.nptlsmp' running, the relevant configuration file is /boot/config-2.4.22-1.2115.nptlsmp. Some distributions have a patch to the kernel that makes it export its configuration file at run-time via /proc/config.gz or some such. Q2: After copying the right config file, when running 'make menuconfig' do I get the correct params as set in the selected config file? You are supposed to, yes, but in the case where you start from an earlier configuration file, it is recommended to use 'make oldconfig' first, answer its questions, and then run 'make menuconfig' and change anything you want. 'make oldconfig' is specifically targeted for the situation where you have an existing configuration file that you want to use as a basis for the current kernel configuration. Q3: Does running 'make mrproper' delete the config file? YES, as well as any other .config* file. You can read the Makefile to see what exactly it does. Q4: Is is important if I compile the encryption as a module or into the kernel? Barring special circumstances, no. Cheers, Muli -- Muli Ben-Yehuda http://www.mulix.org | http://mulix.livejournal.com/ signature.asc _ The new MSN 8: advanced junk mail protection and 2 months FREE* http://join.msn.com/?page=features/junkmail = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
On Wed, Mar 10, 2004 at 09:01:55AM +0200, Yedidyah Bar-David wrote: On Tue, Mar 09, 2004 at 11:14:31PM +0200, Shachar Shemesh wrote: Yedidyah Bar-David wrote: I copied configs/kernel-2.4.22-i686.config to .config, ran 'make menuconfig', immediately selected exit, yes for save. diff between the original conf and the new .config is 2241 lines. Bad enough, you say? Well, now I ran again menuconfig, now entered the first submenu (Code maturity level options), immediately exit, exit, yes for save. There is a diff (although small) between .config between the first run and the second! You are asking 'make menuconfig' to do something it is not designed nor meant to do. That's why we have 'make oldconfig', as Shachar said. try running make oldconfig instead. This asks you (in text mode) only those questions which have no answer in the config file. See how many questions you get. None! Ok, that means that all options are accounted for. How big is the diff between .config files? are there substantial differences? While I had no intentions of starting a long discussion, and studying the details of the config system, I will try to give my own guess: These config files are hand-edited. Problem. You are not supposed to do that. ALWAYS run 'make oldconfig' after hand editing a .config. a .config generated by oldconfig is by definition valid - a .config edited by hand by you isn't. Can you repeat your experiments with 'make oldconfig' rather than 'make menuconfig', and then tell us if there's a problem? Thanks, Muli -- Muli Ben-Yehuda http://www.mulix.org | http://mulix.livejournal.com/ signature.asc Description: Digital signature
Re: kernel compilation
On Wed, Mar 10, 2004 at 10:20:37AM +0200, Muli Ben-Yehuda wrote: On Wed, Mar 10, 2004 at 09:01:55AM +0200, Yedidyah Bar-David wrote: On Tue, Mar 09, 2004 at 11:14:31PM +0200, Shachar Shemesh wrote: Yedidyah Bar-David wrote: [snip] You are asking 'make menuconfig' to do something it is not designed nor meant to do. That's why we have 'make oldconfig', as Shachar said. Well, this isn't written in README. README says: If you want to carry your existing configuration to a new version with minimal work, use make oldconfig, which will only ask you for the answers to new questions. But it doesn't say that's the only way, only the minimal work one. My old understanding was that all the *config targets do that, but oldconfig only asks new questions, while config asks all, and others are too-interactive if that's all you want (and don't make it easy to know what questions were added). Thanks for the advice, though. try running make oldconfig instead. This asks you (in text mode) only those questions which have no answer in the config file. See how many questions you get. None! Ok, that means that all options are accounted for. How big is the diff between .config files? are there substantial differences? Very well. % cp configs/kernel-2.4.22-i686.config .config % make oldconfig (lots of output, no questions) % diff configs/kernel-2.4.22-i686.config .config | wc 21935219 47449 % diff configs/kernel-2.4.22-i686.config .config | grep -v '#' | grep CONFIG | wc 11852370 28945 % grep -v '#' .config | grep CONFIG | sort config-sorted % grep -v '#' configs/kernel-2.4.22-i686.config | grep CONFIG | sort fedora-config-sorted % diff config-sorted fedora-config-sorted | wc 448 7258182 % diff config-sorted fedora-config-sorted | grep CONFIG | wc 277 5546689 % diff config-sorted fedora-config-sorted | grep CONFIG | grep '^' | wc 69 1381732 % diff config-sorted fedora-config-sorted | grep CONFIG | grep '^' | wc 208 4164957 That's 277 actual changes. Furthermore, % cp .config config-after-oldconfig % make oldconfig (again - lots of output, no questions) % diff .config config-after-oldconfig 98a99,100 CONFIG_SHARE_RUNQUEUE=y CONFIG_MAX_NR_SIBLINGS=2 136a139 # CONFIG_HOTPLUG_PCI_IBM is not set % grep CONFIG_SHARE_RUNQUEUE .config configs/kernel-2.4.22-i686.config config-after-oldconfig config-after-oldconfig:CONFIG_SHARE_RUNQUEUE=y % grep CONFIG_MAX_NR_SIBLINGS .config configs/kernel-2.4.22-i686.config config-after-oldconfig config-after-oldconfig:CONFIG_MAX_NR_SIBLINGS=2 Now I became curious. I downloaded vanilla 2.4.22, and did the same: % cp ../linux-2.4.22-1.2115.nptl/configs/kernel-2.4.22-i686.config .config % make oldconfig (again no questions) % cp .config config-after-oldconfig % make oldconfig (no questions) % diff .config config-after-oldconfig 106c106 # CONFIG_HOTPLUG_PCI_ACPI is not set --- # CONFIG_HOTPLUG_PCI_IBM is not set What do you say about that? Bug in oldconfig? Maybe. Only in Fedora's kernel sources? Not really - I would've expect this diff to be really empty, but it's of course way better. Just to make sure, I do the same on vanilla with my own 2.4.22 config: % cp ~/stuff/kernel/configs/config-2.4.22-net3 .config % make oldconfig (one question, about some driver, which proves that Muli is right, and menuconfig (with which I created my config) didn't answer all the questions) % cp .config config-after-oldconfig % make oldconfig (no questions) % diff .config config-after-oldconfig (empty) While I had no intentions of starting a long discussion, and studying the details of the config system, I will try to give my own guess: These config files are hand-edited. Problem. You are not supposed to do that. ALWAYS run 'make oldconfig' after hand editing a .config. a .config generated by oldconfig is by definition valid - a .config edited by hand by you isn't. I didn't say _I_ hand-edited it. I'd never do. I only said my *guess* is that Fedora's kernel-maintainer did. My hint was the doubled line. -- Didi Can you repeat your experiments with 'make oldconfig' rather than 'make menuconfig', and then tell us if there's a problem? -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
On Wed, Mar 10, 2004 at 11:33:51AM +0200, Yedidyah Bar-David wrote: What do you say about that? Bug in oldconfig? Maybe. Only in Fedora's kernel sources? Not really - I would've expect this diff to be really empty, but it's of course way better. Possibly, I don't know. Thanks for checking. It would be interesting to try the same thing with Fedora's 2.6 kernel and see what happens. In 2.6 the build system and the configuration system (kbuild and kconfig) were both significantly changed, as was mentioned up-thread. Cheers, Muli -- Muli Ben-Yehuda http://www.mulix.org | http://mulix.livejournal.com/ signature.asc Description: Digital signature
kernel compilation
Hi, I have a Linux RH9, and I want to compile the kernel to add support for encrypted file system (using the loop device). My question is: How can I set all the parameters compiled in my current kernel version (2.4.20-8) which came with the distribution? (This way I will only change the encryption part and not all the kernel params) Another question: What patch do I have to apply in order to get the encryption working? I googled for it, but there is so much data I got confused. Currently, when I perform the command losetup -e aes-256 /dev/loop0 .crypto I get the error: ioctl: LOOP_SET_STATUS: Invalid argument. I understood that this means that the kernel does not support encryption. Correct me if I'm wrong. Waiting for your replies, David. _ Protect your PC - get McAfee.com VirusScan Online http://clinic.mcafee.com/clinic/ibuy/campaign.asp?cid=3963 = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
On Tuesday 09 March 2004 14:41, David Sapir wrote: I have a Linux RH9, and I want to compile the kernel to add support for encrypted file system (using the loop device). My question is: How can I set all the parameters compiled in my current kernel version (2.4.20-8) which came with the distribution? (This way I will only change the encryption part and not all the kernel params) Install the RedHat kernel source RPM (no, not SRPM, there's an RPM of the kernel sources), then go to /usr/src/linux-version/configs and you'll find the configuration files used to compile the RedHat supplied kernel. There are several config files suitable for SMP/UP, BIGMEM, no BIGMEM etc etc., so just cp the right one as .config and there you go. Another question: What patch do I have to apply in order to get the encryption working? I googled for it, but there is so much data I got confused. Currently, when I perform the command losetup -e aes-256 /dev/loop0 .crypto I get the error: ioctl: LOOP_SET_STATUS: Invalid argument. I understood that this means that the kernel does not support encryption. Correct me if I'm wrong. Are you sure you need a patch and not just to enable support for that specific encryption scheme (AES) during make menuconfig? I wouldn't be surprised if RedHat came with the encryption options turned off because of American EAR regulations. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust (TM) http://www.codefidence.com I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Kernel .config differences? (was: Re: kernel compilation)
Are there still 2241 changes even after you use the options -iEbB, and ignore changes in comment lines and trailing comments? Yedidyah Bar-David wrote: Hi all, I want to point out a question, and might even get an answer. Even if not, whoever regularly compile their own kernels and try different configs, pay attention. [... snipped ...] I copied configs/kernel-2.4.22-i686.config to .config, ran 'make menuconfig', immediately selected exit, yes for save. diff between the original conf and the new .config is 2241 lines. Bad enough, you say? Well, now I ran again menuconfig, now entered the first submenu (Code maturity level options), immediately exit, exit, yes for save. There is a diff (although small) between .config between the first run and the second! --- Omer My opinions, as expressed in this E-mail message, are mine alone. They do not represent the official policy of any organization with which I may be affiliated in any way. WARNING TO SPAMMERS: at http://www.zak.co.il/spamwarning.html = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
Gilad Ben-Yossef wrote: Install the RedHat kernel source RPM (no, not SRPM, there's an RPM of the kernel sources), then go to /usr/src/linux-version/configs and you'll find the configuration files used to compile the RedHat supplied kernel. There are several config files suitable for SMP/UP, BIGMEM, no BIGMEM etc etc., so just cp the right one as .config and there you go. Wouldn't he rather get the SRPM and run rpmbuild? What is in the kernel source RPM anyways? I was under the impression it contained an almost vanilla kernel (as opposed to the SRPM, which is heavilly patched). Where is the SRPM for the kernel source rpm? :-) -- Shachar Shemesh Lingnu Open Systems Consulting http://www.lingnu.com/ = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
Hi Gilad, Thanks for your answer. I found the config files you mentioned. I will try to just enable encryption compilation. I would also like to state that it is the first time I'm trying to compile the kernel, so forgive me if my questions seem stupid or obvious. Q1: How do I know which of the config files were used to compile my kernel (the original distro kernel)? Q2: After copying the right config file, when running 'make menuconfig' do I get the correct params as set in the selected config file? Q3: Does running 'make mrproper' delete the config file? Q4: Is is important if I compile the encryption as a module or into the kernel? Again, thanks for your patience, David. From: Gilad Ben-Yossef [EMAIL PROTECTED] To: David Sapir [EMAIL PROTECTED], [EMAIL PROTECTED] Subject: Re: kernel compilation Date: Tue, 9 Mar 2004 15:05:08 +0200 On Tuesday 09 March 2004 14:41, David Sapir wrote: I have a Linux RH9, and I want to compile the kernel to add support for encrypted file system (using the loop device). My question is: How can I set all the parameters compiled in my current kernel version (2.4.20-8) which came with the distribution? (This way I will only change the encryption part and not all the kernel params) Install the RedHat kernel source RPM (no, not SRPM, there's an RPM of the kernel sources), then go to /usr/src/linux-version/configs and you'll find the configuration files used to compile the RedHat supplied kernel. There are several config files suitable for SMP/UP, BIGMEM, no BIGMEM etc etc., so just cp the right one as .config and there you go. Another question: What patch do I have to apply in order to get the encryption working? I googled for it, but there is so much data I got confused. Currently, when I perform the command losetup -e aes-256 /dev/loop0 .crypto I get the error: ioctl: LOOP_SET_STATUS: Invalid argument. I understood that this means that the kernel does not support encryption. Correct me if I'm wrong. Are you sure you need a patch and not just to enable support for that specific encryption scheme (AES) during make menuconfig? I wouldn't be surprised if RedHat came with the encryption options turned off because of American EAR regulations. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust (TM) http://www.codefidence.com I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie _ STOP MORE SPAM with the new MSN 8 and get 2 months FREE* http://join.msn.com/?page=features/junkmail = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
On Tue, 9 Mar 2004, David Sapir wrote: Thanks for your answer. I found the config files you mentioned. I will try to just enable encryption compilation. I would also like to state that it is the first time I'm trying to compile the kernel, so forgive me if my questions seem stupid or obvious. before asking questions, i suggest you look at something on the net that explains the compilation process. i just recently gave a lecture about this in haifux - check the slides at http://www.haifux.org/lectures/88-sil/ as well as the earlier lecture at http://www.haifux.org/lectures/86-sil/kernel-modules-drivers/ this _might_ help you get started. hope this helps (and sorry for the shameless plug). -- guy For world domination - press 1, or dial 0, and please hold, for the creator. -- nob o. dy = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
Guy, Not only I looked at your slides (I keep them as reference), but I attended your lecture (which I enjoyed very much). This gave me the push to take the encryption assignment and to try compiling the kernel myself. I know compiling the kernet is very easy *if you have done it before*. As I stated, it's the first time I'm trying to compile the kernel, and things are not as easy as it seemed to me on your lecture. To complete my knowledge, I googled. Now after this long speech, can you please answer my questions? To remind you, they were: Q1: How do I know which of the config files were used to compile my kernel (the original distro kernel)? Q2: After copying the right config file, when running 'make menuconfig' do I get the correct params as set in the selected config file? Q3: Does running 'make mrproper' delete the config file? Q4: Is is important if I compile the encryption as a module or into the kernel? Thanks, David. From: guy keren [EMAIL PROTECTED] To: David Sapir [EMAIL PROTECTED] CC: [EMAIL PROTECTED] Subject: Re: kernel compilation Date: Tue, 9 Mar 2004 16:16:57 +0200 (EET) On Tue, 9 Mar 2004, David Sapir wrote: Thanks for your answer. I found the config files you mentioned. I will try to just enable encryption compilation. I would also like to state that it is the first time I'm trying to compile the kernel, so forgive me if my questions seem stupid or obvious. before asking questions, i suggest you look at something on the net that explains the compilation process. i just recently gave a lecture about this in haifux - check the slides at http://www.haifux.org/lectures/88-sil/ as well as the earlier lecture at http://www.haifux.org/lectures/86-sil/kernel-modules-drivers/ this _might_ help you get started. hope this helps (and sorry for the shameless plug). -- guy For world domination - press 1, or dial 0, and please hold, for the creator. -- nob o. dy _ Add photos to your messages with MSN 8. Get 2 months FREE*. http://join.msn.com/?page=features/featuredemail = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
On Tue, Mar 09, 2004 at 04:34:53PM +0200, David Sapir wrote: How do I know which of the config files were used to compile my kernel (the original distro kernel)? Usually, there's correspondence between the kernel version of the running kernel (try 'uname -a' to see it) and the config file's name. For example, on the machine that has kernel '2.4.22-1.2115.nptlsmp' running, the relevant configuration file is /boot/config-2.4.22-1.2115.nptlsmp. Some distributions have a patch to the kernel that makes it export its configuration file at run-time via /proc/config.gz or some such. Q2: After copying the right config file, when running 'make menuconfig' do I get the correct params as set in the selected config file? You are supposed to, yes, but in the case where you start from an earlier configuration file, it is recommended to use 'make oldconfig' first, answer its questions, and then run 'make menuconfig' and change anything you want. 'make oldconfig' is specifically targeted for the situation where you have an existing configuration file that you want to use as a basis for the current kernel configuration. Q3: Does running 'make mrproper' delete the config file? YES, as well as any other .config* file. You can read the Makefile to see what exactly it does. Q4: Is is important if I compile the encryption as a module or into the kernel? Barring special circumstances, no. Cheers, Muli -- Muli Ben-Yehuda http://www.mulix.org | http://mulix.livejournal.com/ signature.asc Description: Digital signature
Re: kernel compilation
On Tuesday 09 March 2004 16:34, David Sapir wrote: Guy, Q1: How do I know which of the config files were used to compile my kernel (the original distro kernel)? SMP - if you have more than one CPU i386/i686/athlon - depends on your CPU type (cat /proc/cpuinfo if you don't know) bigmem - if you have more than 4GB ram (read: rpm -qip kernelbigmem.rpm) debug - only if you installed it manually. So probably, your config is (substitute your processor): kernel-2.4.20-i586.config Q2: After copying the right config file, when running 'make menuconfig' do I get the correct params as set in the selected config file? Yes. But you can always load them explicitly from the UI. Q3: Does running 'make mrproper' delete the config file? Yes. Run it before copying the config file. If this is your first kernel compile, than you *must* run this (at least on RedHat systems, but it's a good idea anyway). Q4: Is is important if I compile the encryption as a module or into the kernel? Why don't you check both options and let us all know your conclusions? -- Oron Peled Voice/Fax: +972-4-8228492 [EMAIL PROTECTED] http://www.actcom.co.il/~oron First they ignore you, then they laugh at you, then they fight you, then you win. -- Gandhi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
On Tuesday 09 March 2004 15:31, Shachar Shemesh wrote: Gilad Ben-Yossef wrote: Install the RedHat kernel source RPM (no, not SRPM, there's an RPM of the kernel sources), then go to /usr/src/linux-version/configs and you'll find the configuration files used to compile the RedHat supplied kernel. There are several config files suitable for SMP/UP, BIGMEM, no BIGMEM etc etc., so just cp the right one as .config and there you go. Wouldn't he rather get the SRPM and run rpmbuild? Since what he wants to do is build a vanilla kernel, but use the same configuration file as RedHat used to save time and hassle (I do it it myself quite often :-), then no - he wants the kernel source RPM which holds those config files. What is in the kernel source RPM anyways? I was under the impression it contained an almost vanilla kernel (as opposed to the SRPM, which is heavilly patched). No, it's a /usr/src/linux-version that contain a Redhat patched version of the kernel. Usefull if you want to debug the kernel, build some (mostly badly written) modules that require kernel sources to build etc. Where is the SRPM for the kernel source rpm? :-) A single SRPM can generate several RPMs. In this case the relvant SRPM generates both the kernel RPM, the kernel headers RPM and the kernel source RPM. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust (tm) http://www.codefidence.com I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
On Tue, Mar 09, 2004 at 07:08:39PM +0200, Gilad Ben-Yossef wrote: No, it's a /usr/src/linux-version that contain a Redhat patched version of the kernel. Usefull if you want to debug the kernel, build some (mostly badly written) modules that require kernel sources to build etc. Hmpf, pet peeve alert. It's modules that do NOT require the kernel sources to be built that are badly written. External modules *must* be built via the kernel's build system, and for that you need the kernel sources. Any other method of building a kernel module is unsafe and bound to bite the builder in the a** - and rightly so... Cheeers, Muli -- Muli Ben-Yehuda http://www.mulix.org | http://mulix.livejournal.com/ signature.asc Description: Digital signature
Re: kernel compilation
Muli Ben-Yehuda wrote: On Tue, Mar 09, 2004 at 07:08:39PM +0200, Gilad Ben-Yossef wrote: No, it's a /usr/src/linux-version that contain a Redhat patched version of the kernel. Usefull if you want to debug the kernel, build some (mostly badly written) modules that require kernel sources to build etc. Hmpf, pet peeve alert. It's modules that do NOT require the kernel sources to be built that are badly written. External modules *must* be built via the kernel's build system, and for that you need the kernel sources. Any other method of building a kernel module is unsafe and bound to bite the builder in the a** - and rightly so... Cheeers, Muli I think you misunderstood Gilad (though you may have not, in which case I misunderstood you) (it is also I misunderstood Gilad, so please, anyone, answer with what you meant). What I think Gilad meant is the distinction between modules requiring the kernel sources, and modules requring just the kernel HEADERS. When you are not working with kernels you compiled yourself (such as distro kernels), not having to have a kernel compilation tree in order to compile a module is a major plus. If that is also what you meant, then I would like to know why you think these modules that can make do with the headers are badly written? -- Shachar Shemesh Lingnu Open Systems Consulting http://www.lingnu.com/ = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
On Tue, Mar 09, 2004 at 03:37:18PM +0200, David Sapir wrote: Hi Gilad, Thanks for your answer. I found the config files you mentioned. I will try to just enable encryption compilation. I would also like to state that it is the first time I'm trying to compile the kernel, so forgive me if my questions seem stupid or obvious. Q1: How do I know which of the config files were used to compile my kernel (the original distro kernel)? See /boot/config-* (Now why won't RH apply the /proc/config.gz patch ?) Q2: After copying the right config file, when running 'make menuconfig' do I get the correct params as set in the selected config file? Yes Q3: Does running 'make mrproper' delete the config file? Yes Q4: Is is important if I compile the encryption as a module or into the kernel? If your encrypted FS is the root FS then you'll need it built-in. If not: you probably don't need it compiled it. -- Tzafrir Cohen +---+ http://www.technion.ac.il/~tzafrir/ |vim is a mutt's best friend| mailto:[EMAIL PROTECTED] +---+ = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
On Tue, Mar 09, 2004 at 07:37:34PM +0200, Shachar Shemesh wrote: I think you misunderstood Gilad (though you may have not, in which case I misunderstood you) (it is also I misunderstood Gilad, so please, anyone, answer with what you meant). Look, it's all just one big misunderstanding. Can't we all just get along? Seriously, we'll have to wait for Gilad to tell us what he meant. What I think Gilad meant is the distinction between modules requiring the kernel sources, and modules requring just the kernel HEADERS. When you are not working with kernels you compiled yourself (such as distro kernels), not having to have a kernel compilation tree in order to compile a module is a major plus. Until the addition of object dir is seperate from source dir support to kbuild in late 2.5 / early 2.6, for building external modules via kbuild you needed headers + Makefiles + some generated files, which lived in the source directory. If you need that much, why not have the source there as well and be done with it. If that is also what you meant, then I would like to know why you think these modules that can make do with the headers are badly written? replace 'headers' with 'headers + Makefiles + kernel build generated files' and I have no beef with them. Cheers, Muli -- Muli Ben-Yehuda http://www.mulix.org | http://mulix.livejournal.com/ signature.asc Description: Digital signature
Re: kernel compilation
On Tue, Mar 09, 2004 at 07:46:50PM +0200, Tzafrir Cohen wrote: If your encrypted FS is the root FS then you'll need it built-in. Either built-in or you need to build an initrd, which if you're using the RH config, you need anyway, since IDE and SCSI are both modular by default. Cheers, Muli -- Muli Ben-Yehuda http://www.mulix.org | http://mulix.livejournal.com/ signature.asc Description: Digital signature
Re: kernel compilation
Hi all, I want to point out a question, and might even get an answer. Even if not, whoever regularly compile their own kernels and try different configs, pay attention. I noticed this a few days ago, on Fedora Core 1. Since I currently do not have access to such a machine, I did not try it on such, but on RedHat 7.3. But - with Fedora's kernel sources, and gcc 3.2.0 (which shouldn't make a difference - I only ran menuconfig, which is hopefully a trivial program, independent of a specific compiler. I did this since make menuconfig wanted gcc32, and that's the closest I had arround - they have a patched 3.2.2, IIRC). I copied configs/kernel-2.4.22-i686.config to .config, ran 'make menuconfig', immediately selected exit, yes for save. diff between the original conf and the new .config is 2241 lines. Bad enough, you say? Well, now I ran again menuconfig, now entered the first submenu (Code maturity level options), immediately exit, exit, yes for save. There is a diff (although small) between .config between the first run and the second! Now, one might explain that since RedHat (Fedora?) has tons of patches to the kernel, also the possible options are different, and although I am _not_ doing this on vanilla kernel (which will probably be different in what features it has), but on Fedora's, there might be something I am missing which explains this. But why would there be a diff between after the first run and after the second run is beyond me. Now, I do not want to spread smoke (is this how you say it in English?), but I am pretty certain that on the real Fedora, for some weird reason, it was even worse - there were actual, real, noticable differences between the configs, that I could see. I did not go over this 2200 lines diff, but I do see signs of such things - e.g. configs/kernel-2.4.22-i686.config has *twice* this line CONFIG_HIGHMEM4G=y while the output .config has it once, and also has CONFIG_HIGHMEM=y which does not appear at all in configs/kernel-2.4.22-i686.config. I do not see immediately how this specific diff can cause real changes in the kernel, but it's also not at all obvious that it doesn't. Any idea? Anyway, all of this will hopefully be better in 2.6, where (so I heard, I did not read it myself) the kernel config system was heavily patched (somewhat rewritten? I do know ESR's CML2 didn't make it in). For the record - while I do not pretend to know deeply how all this config stuff works, and do not give any guarantees to the following, I will say how I work. At home, I have a quite-tuned config - I once skimmed over the entire menuconfig, carefully chosen what I want (and the hardware I have), and since then I copy over the config to every new version I compile, adding features and drivers as I need/want. But I know that the base was good, and the changes are incremental. When I started working here (in tau), 3+ years ago, I intended to do mostly the same, but did not know what hardware there is, and also did not want to compile a new kernel (or driver) every few weeks. So I started with RedHat's config on a vanilla kernel, and continued the same as at home since then - not following RedHat but using the last good config. So I got a base config of many drivers, which take around twice+ to compile with all those modules, and am mostly happy. After this experience with Fedora a few days ago, I can't anymore recommend their config whole-heartedly. I also stopped treating it as passive, trivial data. Unless someone here will have a good explanation. I will personally continue using my configs, and do not mind posting them if anybody wants them. -- Didi On Tue, Mar 09, 2004 at 07:08:39PM +0200, Gilad Ben-Yossef wrote: On Tuesday 09 March 2004 15:31, Shachar Shemesh wrote: Gilad Ben-Yossef wrote: Install the RedHat kernel source RPM (no, not SRPM, there's an RPM of the kernel sources), then go to /usr/src/linux-version/configs and you'll find the configuration files used to compile the RedHat supplied kernel. There are several config files suitable for SMP/UP, BIGMEM, no BIGMEM etc etc., so just cp the right one as .config and there you go. Wouldn't he rather get the SRPM and run rpmbuild? Since what he wants to do is build a vanilla kernel, but use the same configuration file as RedHat used to save time and hassle (I do it it myself quite often :-), then no - he wants the kernel source RPM which holds those config files. What is in the kernel source RPM anyways? I was under the impression it contained an almost vanilla kernel (as opposed to the SRPM, which is heavilly patched). No, it's a /usr/src/linux-version that contain a Redhat patched version of the kernel. Usefull if you want to debug the kernel, build some (mostly badly written) modules that require kernel sources to build etc. Where is the SRPM for the kernel source rpm? :-) A single SRPM can generate several RPMs. In this case the relvant SRPM generates both the kernel
Re: kernel compilation
Yedidyah Bar-David wrote: I copied configs/kernel-2.4.22-i686.config to .config, ran 'make menuconfig', immediately selected exit, yes for save. diff between the original conf and the new .config is 2241 lines. Bad enough, you say? Well, now I ran again menuconfig, now entered the first submenu (Code maturity level options), immediately exit, exit, yes for save. There is a diff (although small) between .config between the first run and the second! try running make oldconfig instead. This asks you (in text mode) only those questions which have no answer in the config file. See how many questions you get. -- Shachar Shemesh Lingnu Open Systems Consulting http://www.lingnu.com/ = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Kernel .config differences? (was: Re: kernel compilation)
On Tue, Mar 09, 2004 at 11:07:33PM +0200, Omer Zak wrote: Are there still 2241 changes even after you use the options -iEbB, and ignore changes in comment lines and trailing comments? My diff doesn't know about 'E'. % diff -ibB configs/kernel-2.4.22-i686.config .config | grep -v '#' | grep CONFIG | wc 11702340 28448 -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation
On Tue, Mar 09, 2004 at 11:14:31PM +0200, Shachar Shemesh wrote: Yedidyah Bar-David wrote: I copied configs/kernel-2.4.22-i686.config to .config, ran 'make menuconfig', immediately selected exit, yes for save. diff between the original conf and the new .config is 2241 lines. Bad enough, you say? Well, now I ran again menuconfig, now entered the first submenu (Code maturity level options), immediately exit, exit, yes for save. There is a diff (although small) between .config between the first run and the second! try running make oldconfig instead. This asks you (in text mode) only those questions which have no answer in the config file. See how many questions you get. None! While I had no intentions of starting a long discussion, and studying the details of the config system, I will try to give my own guess: These config files are hand-edited. They are based on real (where real means the output of one of the *config targets) files, but with additions. The obvious example I already pointed out is the duplicate line 'CONFIG_HIGHMEM4G=y'. While I did not study the sources, I don't find it logical that a script will output such a line twice, especially not sometimes (as I said, in my output it doesn't). If my guess is correct, I find it very bad, on the part of Fedora. Even giving a real conf isn't bullet-proof, and that's why this system was deeply fixed in 2.6. But hand-editing? That's too far, I think. And with 2000 (or 1000, if you want) diffs? They even did not try to make it seem ok. I have no other explanation. Now, I thank you and Omer, but I do not want to start an investigation. Either someone (Muli? Gilad?) has an answer, which I will be glad to hear, or someone is enough interested that he will play with this himself and tell us. Personally, I already wrote what I do, and have no intentions to change. Just to make sure, I now ran menuconfig on my latest config, entered many submenus and exited without changing, and the diff was, as expected, empty. That's what it sould be, as far as I know. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
alternative CC for kernel compilation on debian-unstable
Hi
I've just failed to build a kernel and figured out that using gcc 3.3 is
not the best way to get my kernel compiled.
I recall that the issue of problem overriding the version of gcc when
using make-kpkg .
I see the following in /usr/bin/make-kpkg:
my $cc = $ENV{'CC'};
$cc =~ s/^\s+//g;
$cc =~ s/\s+$//g;
$cc .= ;
if ($cc eq ' ') { $cc = $cross_compile; $cc .= gcc ; };
#$cc .= -D__KERNEL__ -I;
#chomp($cc .= `pwd`);
#$cc .= /include;
# For some reason, this causes all modules to fail
#$command .= CC=\$cc\ HOSTCC=\$hostcc\;
$command .= $Targets;
# print STDERR $command;
exec $command;
So the following trivial change seems to do the trick:
--- /tmp/make-kpkg_orig 2003-06-12 02:08:45.0 +0300
+++ /usr/bin/make-kpkg 2003-06-12 02:09:21.0 +0300
@@ -962,7 +962,7 @@
#$cc .= /include;
# For some reason, this causes all modules to fail
- #$command .= CC=\$cc\ HOSTCC=\$hostcc\;
+ $command .= CC=\$cc\ HOSTCC=\$hostcc\;
$command .= $Targets;
# print STDERR $command;
exec $command;
But this looked too simple to last that much unfixed, and indeed
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=196197 suggests
that this is not the way to go.
Anyway, I have just had the building with gcc-2.95 fail at the same
file. The problem seems to be in the uml patch :-(
--
Tzafrir Cohen +---+
http://www.technion.ac.il/~tzafrir/ |vim is a mutt's best friend|
mailto:[EMAIL PROTECTED] +---+
=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word unsubscribe in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation time
It doesn't sound normal to me. try to do /usr/bin/time make bzlilo log and see the %CPU. If it not close to 100%, then You either run other things that take CPU time, or you swap alot. Try running it without Xwindows. Last time I compiled a kernel (2.2.5) on my old 486-66 with 64MB, it was 52:45.21elapsed 97%CPU. Of course, you probably have another set of drivers and stuff, and the 64MB do help. Also, try running top on another VC, and see how much swapping you do. You can always do "make -j 2". This will probably help you do it slightly faster. Good rule - run make -j N+1 where N is number of CPU's in your computer. Meir = To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re[2]: kernel compilation time
Meir Litmanovich [EMAIL PROTECTED] wrote: You can always do "make -j 2". This will probably help you do it slightly faster. Good rule - run make -j N+1 where N is number of CPU's in your computer. I don't understand the meaning of that "+1" (unless this is a way to concur with another CPU-intensive process running simultaneously on the same box - but then why not just SIGSTOP or renice it for awhile??). And, definitely, parallelizing kernel make on a machine with only 16MB of RAM installed is a bad idea anyway, no matter how many CPUs it has. Regards, Evgeny -- / Evgeny Stambulchik [EMAIL PROTECTED] \ / Plasma Laboratory, Weizmann Institute of Science, Israel \ \ | Phone : (972)8-934-3610 == | == FAX : (972)8-934-3491 | | | URL :http://plasma-gate.weizmann.ac.il/~fnevgeny/ | | | Finger for PGP key =+ | |__| = To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re[4]: kernel compilation time
Meir Litmanovich [EMAIL PROTECTED] wrote: I don't understand the meaning of that "+1" (unless this is a way to concur with another CPU-intensive process running simultaneously on the same box - but then why not just SIGSTOP or renice it for awhile??). Oh, not. Just imagine your computer does nothing except for kernel compilation (that's the situation, right ? ). Then you really don't want to wait for I/O on this machine. N+1 ensures that in time one gcc would stuck waiting for I/O, other will effectively run. But why would one wait for I/O during compilation on an otherwise idle machine? Only if one has an AT disk or something similar by I/O performance, like NFS over very busy ethernet. Or, if the box is heavily swapping - but then an extra compilation process wouldn't help :). But in normal situations this shouldn't happen, IMO. At least, I've never seen any difference between compile times of make -j N and make -j N+1 (or, for that matter, N+k for any k 0 - till the RAM is exhausted). Tried on several uni-processor and SMP boxes. Regards, Evgeny -- / Evgeny Stambulchik [EMAIL PROTECTED] \ / Plasma Laboratory, Weizmann Institute of Science, Israel \ \ | Phone : (972)8-934-3610 == | == FAX : (972)8-934-3491 | | | URL :http://plasma-gate.weizmann.ac.il/~fnevgeny/ | | | Finger for PGP key =+ | |__| = To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
kernel compilation time
I'm compiling a 2.2.5 kernel (that comes with RH6) on a DX4 with 16MB. I made dep and clean, and now running bzlilo. The last part (make bzlilo) has been running since 12:00, which makes it a total of over 7(!) hours and counting. Is this a normal speed? Is there any way to find out what slowing the computer down? (unless it's normal, of course). Almost all the services are down, nothing is running in the background. - Aviram Jenik "Addicted to Chaos" - Today's quote: I was recently on a tour of Latin America, and the only regret I have was that I didn't study Latin harder in school so I could converse with those people. - J. Danforth Quayle, quoted in "Time", 8 May 1989 = To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation time
Aviram. Yes, it's normal. I still got a 486 DX-2 which is found not near me and I upgrade it's kernel.. But the best way to do it is to compile the kernel on stronger machine, and copy the results file (vmlinuz, system.map) to the 486, and update the lilo... Hetz Aviram Jenik wrote: I'm compiling a 2.2.5 kernel (that comes with RH6) on a DX4 with 16MB. I made dep and clean, and now running bzlilo. The last part (make bzlilo) has been running since 12:00, which makes it a total of over 7(!) hours and counting. Is this a normal speed? Is there any way to find out what slowing the computer down? (unless it's normal, of course). Almost all the services are down, nothing is running in the background. - Aviram Jenik "Addicted to Chaos" - Today's quote: I was recently on a tour of Latin America, and the only regret I have was that I didn't study Latin harder in school so I could converse with those people. - J. Danforth Quayle, quoted in "Time", 8 May 1989 = To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED] -- Hetz Ben Hamo - Sys. Admin. - Intercomp [EMAIL PROTECTED] --- Redmond, you have a problem.. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel compilation time
Hi Aviram Jenik wrote: I'm compiling a 2.2.5 kernel (that comes with RH6) on a DX4 with 16MB. I made dep and clean, and now running bzlilo. The last part (make bzlilo) has been running since 12:00, which makes it a total of over 7(!) hours and counting. Is this a normal speed? Is there any way to find out what slowing the computer down? (unless it's normal, of course). Almost all the services are down, nothing is running in the background. It doesn't sound normal to me. try to do /usr/bin/time make bzlilo log and see the %CPU. If it not close to 100%, then You either run other things that take CPU time, or you swap alot. Try running it without Xwindows. Last time I compiled a kernel (2.2.5) on my old 486-66 with 64MB, it was 52:45.21elapsed 97%CPU. Of course, you probably have another set of drivers and stuff, and the 64MB do help. Also, try running top on another VC, and see how much swapping you do. The best thing, however, is was Hetz suggested: Compile it on another machine. I recall seeing a web site, where you send a config file, and get in return a kernel. Don't remember URL, sorry. - Aviram Jenik "Addicted to Chaos" - Today's quote: I was recently on a tour of Latin America, and the only regret I have was that I didn't study Latin harder in school so I could converse with those people. - J. Danforth Quayle, quoted in "Time", 8 May 1989 = To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED] = To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
