Re: Unable to compile GCC-4.4.3, Pass 1
On 03/26/2010 11:36 PM, stosss wrote: > > You really have a bad attitude. I wasn't being insulting. You are the > worst one when it comes to be caustic toward any suggestions. Your > reply is always, "the book is clear". You are also quick to say, "you > did something wrong" "you didn't read some part of the book". "if you > read the book you wouldn't have made that mistake" You are the primary > one for so much friction in this project. > > Go learn how to communicate. > Yeah, most of the time, peoples' problems *are* because they don't read the book. Granted, on occasion I *do* jump to that conclusion a bit early and unnecessarily, but past history makes it clear that many people just skip much of the text (and yes, a number of users who have problems have *admitted* to not paying much attention to the text - or frequently people do read but just forget some important part), which is what causes most problems. Oh well, why am I concerned with your opinion of me...my primary interest is in ensuring that the book is both technically and grammatically accurate, as well as understandable for the non-expert...and that's true whether you think so or not. So, sometimes I happen to disagree on what might be "correct" or "understandable"...big deal, nobody's perfect... I'd like to say something like "I don't care what you say so I'm just not responding to further comments" but then I know how much I always just *have* to add further responses (this one being a case in point), so...we'll see how I do... -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
> Oh geez, more "you experts don't comprehend thought processes of mere > mortals" nonsense. How about skipping the generic "you know because > you've done it 1000 times" and just get to the actual point? I don't > need anyone to tell me how important it is to make sure the book is > clear - ensuring that the text is technically correct, and more > importantly understandable, is my primary goal. > > And just for the record, I did also understand it just fine the *first* > time I did it, so don't even think you can give me that "you only know > because you're an expert" crap. > > If there is a specific issue, please make a suggestion. You won't get > anywhere just repeating that "the book is confusing and you non-newbies > just don't comprehend how to make it readable." You really have a bad attitude. I wasn't being insulting. You are the worst one when it comes to be caustic toward any suggestions. Your reply is always, "the book is clear". You are also quick to say, "you did something wrong" "you didn't read some part of the book". "if you read the book you wouldn't have made that mistake" You are the primary one for so much friction in this project. Go learn how to communicate. -- If we can but prevent the government from wasting the labours of the people, under the pretence of taking care of them, they must become happy. - Thomas Jefferson -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
On 03/26/2010 10:56 PM, stosss wrote: > > When I first went through the book it was very confusing. After just > doing it and not knowing if I was doing it right, and having done it > right and having built several systems that work, now the confusion is > gone. > > For first time readers there are several places in the book that are > confusing. Yes there are notes at the beginning of chapters 4, 5 and 6 > that apply to the entire chapter. First time readers who are not as > experienced don't always remember what they read at the beginning of > the chapter when they are trying to get what is in front of them done > right. > > Chris, because you have done this many times it is clear to you. You > also apparently know this stuff well enough that you can't understand > why others have difficulty with it. Can't see the forest because of > the trees. Oh geez, more "you experts don't comprehend thought processes of mere mortals" nonsense. How about skipping the generic "you know because you've done it 1000 times" and just get to the actual point? I don't need anyone to tell me how important it is to make sure the book is clear - ensuring that the text is technically correct, and more importantly understandable, is my primary goal. And just for the record, I did also understand it just fine the *first* time I did it, so don't even think you can give me that "you only know because you're an expert" crap. If there is a specific issue, please make a suggestion. You won't get anywhere just repeating that "the book is confusing and you non-newbies just don't comprehend how to make it readable." -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
stosss wrote: > I would like to see this project and the book improved. Suggestions on wording are welcome. Patches to the xml are even more welcome, even if they are not incorporated directly in the book. They help to highlight exactly what changes are being proposed. > Besides how many people are actually involved with this project? For LFS, primarily just Matt and I are making changes to the most recent versions of the book. Look at the change log. Every change says who made the edit. There are many over the years though, and we try to acknowledge those that have made suggestions that have been incorporated. -- Bruce -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
On Fri, Mar 26, 2010 at 10:12 PM, Chris Staub wrote: > On 03/26/2010 10:00 PM, Chris Staub wrote: >> On 03/26/2010 09:22 PM, Simon Geard wrote: >>> >>> "The GCC documentation recommends building GCC outside >>> of it's source directory in a dedicated build directory." >>> >>> >>> To me, replacing "the" with "it's" makes is much clearer that it refers >>> to GCC's source directory, not to the higher-level one containing the >>> tarballs. >>> >>> Simon. >>> >> >> Well, I still don't agree that the way it is now is confusing (unless >> you're reading too much into it), but I guess "its" (and that's "its" >> not "it's") is technically a bit more accurate anyway, so may as well >> change it. > > Of course, the same would apply to Binutils and Glibc as well, and > arguably E2fsprogs... When I first went through the book it was very confusing. After just doing it and not knowing if I was doing it right, and having done it right and having built several systems that work, now the confusion is gone. For first time readers there are several places in the book that are confusing. Yes there are notes at the beginning of chapters 4, 5 and 6 that apply to the entire chapter. First time readers who are not as experienced don't always remember what they read at the beginning of the chapter when they are trying to get what is in front of them done right. Chris, because you have done this many times it is clear to you. You also apparently know this stuff well enough that you can't understand why others have difficulty with it. Can't see the forest because of the trees. The book is not clear and there are places that do need to be written clearer. This is not an attack on anyone who is connected to what is in the book. It is hard for you who maintain the book to see why and how the new reader can be confused because you know the process so well isn't confusing for you. This is where this project has always had a problem. New readers say the book is confusing and those with much experience at doing such things get defensive and say the book is clear and take issue with any one that says the book is not clear. I would like to see those who know this stuff well and who write the book, be more willing to listen to others who are not as experienced and then figure out how to change the book so it is clear. There have been some that say the book is for experienced readers and not new people and others how point out that some of the command processes in the book actually do cater to new readers and not just those with more experience. I would like to see this project and the book improved. Everyone has to be willing to cooperate before new people that come along and could help might be willing to stick around and help. Besides how many people are actually involved with this project? I don't mean how many people have registered. How many people are doing the work? What will happen to this project when they get tired of doing the work? I see about 3 or 4 people. Two of them actually writing the book. This project has to be a grind at times for them. -- If we can but prevent the government from wasting the labours of the people, under the pretence of taking care of them, they must become happy. - Thomas Jefferson -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
On 03/26/2010 10:00 PM, Chris Staub wrote: > On 03/26/2010 09:22 PM, Simon Geard wrote: >> >> "The GCC documentation recommends building GCC outside >> of it's source directory in a dedicated build directory." >> >> >> To me, replacing "the" with "it's" makes is much clearer that it refers >> to GCC's source directory, not to the higher-level one containing the >> tarballs. >> >> Simon. >> > > Well, I still don't agree that the way it is now is confusing (unless > you're reading too much into it), but I guess "its" (and that's "its" > not "it's") is technically a bit more accurate anyway, so may as well > change it. Of course, the same would apply to Binutils and Glibc as well, and arguably E2fsprogs... -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
On 03/26/2010 09:22 PM, Simon Geard wrote: > > Well, if we're seeing people being confused by what 'source directory' > refers to in this case, it's probably worth trying to clarify it a > little. Even a single-word change like the following might help: > > "The GCC documentation recommends building GCC outside > of it's source directory in a dedicated build directory." > > > To me, replacing "the" with "it's" makes is much clearer that it refers > to GCC's source directory, not to the higher-level one containing the > tarballs. > > Simon. > Well, I still don't agree that the way it is now is confusing (unless you're reading too much into it), but I guess "its" (and that's "its" not "it's") is technically a bit more accurate anyway, so may as well change it. -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
On Fri, 2010-03-26 at 11:33 -0400, Chris Staub wrote: > Yes, this is referring specifically to GCC's source directory, not the > $LFS/sources dir, and a few people have said they were slightly > confused by this as well, but then I think the way it is in the book > is clear enough as it is. I believe LFS' intended audience should be > able to understand that distinction. Well, if we're seeing people being confused by what 'source directory' refers to in this case, it's probably worth trying to clarify it a little. Even a single-word change like the following might help: "The GCC documentation recommends building GCC outside of it's source directory in a dedicated build directory." To me, replacing "the" with "it's" makes is much clearer that it refers to GCC's source directory, not to the higher-level one containing the tarballs. Simon. signature.asc Description: This is a digitally signed message part -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
John Stephens wrote: > Chris, > Thanks for the reply. I've spent a lot of time in chapters 4 and 5. > I'm finding the book (6.6) a little unclear in certain areas which is > probably what is throwing me off to some degree. > > For example, look at 5.5.1, first box. It says: > > tar -jxf ../mpfr-2.4.2.tar.bz2 > mv -v mpfr-2.4.2 mpfr > tar -jxf ../gmp-5.0.0.tar.bz2 > mv -v ../gmp-5.0.0 gmp > > What is the point of reference for "../pkg-name"? If I am in my > /mnt/lfs/sources dir, then the ../ takes me up to /mnt/lfs and the > source tarball is not there. In the "General Instructions" you'll see that, for each package you are going to build, you are to untar the tarball, then cd to the directory which came out of the tarball. Each section associated with a package presumes that you have already done that before it describes the steps to follow. > > I haven't found any reference as to where I should be sitting > (directory wise...) when I issue the commands. Look in the general instructions. Mike -- p="p=%c%s%c;main(){printf(p,34,p,34);}";main(){printf(p,34,p,34);} Oppose globalization and One World Governments like the UN. This message made from 100% recycled bits. You have found the bank of Larn. I speak only for myself, and I am unanimous in that! -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
On Fri, Mar 26, 2010 at 11:33 AM, Chris Staub wrote: > > Yes, this is referring specifically to GCC's source directory, not the > $LFS/sources dir, and a few people have said they were slightly confused > by this as well, but then I think the way it is in the book is clear > enough as it is. > > No, the instructions are correct as they are. Again, you are assumed (as > per the Note on page 5.3) to be doing those commands from within the GCC > source dir, so GMP and MPFR will be unpacked there, and that is where > their unpacked source directories will be. Chris. Thanks. That clears things up for me. Seems every project has its own set of procedures and doc formatting (Imagine that!) and sometimes it takes a bit to get into the same mindset as the dev / writers. I'll give it another go now and see what happens. -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
On 03/26/2010 09:54 AM, John Stephens wrote: > Chris and Phillippe, > > Thanks for the insights. Like I stated before, I was either missing > something conceptually or misinterpreting something. > > I took the instructions "The GCC documentation recommends building GCC > outside of the source directory in a dedicated build directory." to > mean outside of /mnt/lfs/source (i.e. at the /mnt/lfs level) instead > of outside /mnt/lfs/source/gcc-4.4.3 (i.e. at the /mnt/lfs/source > level). Subtle, but important difference. Yes, this is referring specifically to GCC's source directory, not the $LFS/sources dir, and a few people have said they were slightly confused by this as well, but then I think the way it is in the book is clear enough as it is. I believe LFS' intended audience should be able to understand that distinction. Also, the text is really just explaining why you are doing the commands, and the exact commands you are to be doing are right there - if you simply take the information on page 5.3 (which applies to all packages), and just go to each package and run the commands while ignoring the majority of the text (not recommended of course, but possible) then everything will likely work fine. In this case, the comment about the GCC docs recommending it be built in a separate dir is nothing more than explaining the reason for the following command, which is what actually creates that build directory. Technically /mnt/lfs/gcc-build also fits that sentence, as it is also outside the source dir, but then if you've been following the instructions (particularly the oft-repeated note about starting in the GCC source dir) then you simply are not going to create /mnt/lfs/gcc-build. Also, the fact that /mnt/lfs/gcc-build is not correct should become fairly obvious when you simply cannot create is as the lfs user. Many people tend to assume at that point that the book left out a step, and proceed either to chown $LFS, or use sudo to create the build dir in /mnt/lfs. In general the first step whenever you have an error should be to go back and check to see whether you've followed all the instructions, rather than (what most people do) just make assumptions and proceed to do things like force it with root. > This being the case, should the instructions on 5.5.1 (assuming I am > sitting in /mnt/lfs/source/gcc-4.4.3 as instructed): > tar -jxf ../mpfr-2.4.2.tar.bz2 > mv -v mpfr-2.4.2 mpfr > tar -jxf ../gmp-5.0.0.tar.bz2 > mv -v gmp-5.0.0 gmp > > really be: > tar -jxf ../mpfr-2.4.2.tar.bz2 > mv -v ../mpfr-2.4.2 ../mpfr > tar -jxf ../gmp-5.0.0.tar.bz2 > mv -v ../gmp-5.0.0 ../gmp > > Seriously, I am not attempting to be argumentative and again I might > be missing something here. However this and the issue with "source > directory" has caused me a little grief. No, the instructions are correct as they are. Again, you are assumed (as per the Note on page 5.3) to be doing those commands from within the GCC source dir, so GMP and MPFR will be unpacked there, and that is where their unpacked source directories will be. -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Problems with HAL
On 3/25/10, Andrew Benton wrote: > HAL is not part of LFS, you should direct questions about things in BLFS to > blfs-supp...@linuxfromscratch.org I configure HAL with --disable-console-kit \ --disable-policy-kit Otherwise new hal expects them and if they are to be installed, it leads to many dependencies that I am not interested in. Useable documentation for these things might exist, though they are gibberish to me. -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
On 26 March 2010 13:54, John Stephens wrote: > > This being the case, should the instructions on 5.5.1 (assuming I am > sitting in /mnt/lfs/source/gcc-4.4.3 as instructed): > tar -jxf ../mpfr-2.4.2.tar.bz2 > mv -v mpfr-2.4.2 mpfr > tar -jxf ../gmp-5.0.0.tar.bz2 > mv -v gmp-5.0.0 gmp > > really be: > tar -jxf ../mpfr-2.4.2.tar.bz2 > mv -v ../mpfr-2.4.2 ../mpfr > tar -jxf ../gmp-5.0.0.tar.bz2 > mv -v ../gmp-5.0.0 ../gmp > > Seriously, I am not attempting to be argumentative and again I might > be missing something here. However this and the issue with "source > directory" has caused me a little grief. > You need gmp and mpfr as directories within the gcc source. The book's instructions will do that. Please stop top-posting, and trim what you are replying to. Thanks. ĸen -- After tragedy, and farce, "OMG poneys!" -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
Chris and Phillippe, Thanks for the insights. Like I stated before, I was either missing something conceptually or misinterpreting something. I took the instructions "The GCC documentation recommends building GCC outside of the source directory in a dedicated build directory." to mean outside of /mnt/lfs/source (i.e. at the /mnt/lfs level) instead of outside /mnt/lfs/source/gcc-4.4.3 (i.e. at the /mnt/lfs/source level). Subtle, but important difference. This being the case, should the instructions on 5.5.1 (assuming I am sitting in /mnt/lfs/source/gcc-4.4.3 as instructed): tar -jxf ../mpfr-2.4.2.tar.bz2 mv -v mpfr-2.4.2 mpfr tar -jxf ../gmp-5.0.0.tar.bz2 mv -v gmp-5.0.0 gmp really be: tar -jxf ../mpfr-2.4.2.tar.bz2 mv -v ../mpfr-2.4.2 ../mpfr tar -jxf ../gmp-5.0.0.tar.bz2 mv -v ../gmp-5.0.0 ../gmp Seriously, I am not attempting to be argumentative and again I might be missing something here. However this and the issue with "source directory" has caused me a little grief. On Fri, Mar 26, 2010 at 7:18 AM, Chris Staub wrote: > On 03/26/2010 07:04 AM, John Stephens wrote: >> Chris, >> Thanks for the reply. I've spent a lot of time in chapters 4 and 5. >> I'm finding the book (6.6) a little unclear in certain areas which is >> probably what is throwing me off to some degree. >> >> For example, look at 5.5.1, first box. It says: >> >> tar -jxf ../mpfr-2.4.2.tar.bz2 >> mv -v mpfr-2.4.2 mpfr >> tar -jxf ../gmp-5.0.0.tar.bz2 >> mv -v ../gmp-5.0.0 gmp >> >> What is the point of reference for "../pkg-name"? If I am in my >> /mnt/lfs/sources dir, then the ../ takes me up to /mnt/lfs and the >> source tarball is not there. >> >> I haven't found any reference as to where I should be sitting >> (directory wise...) when I issue the commands. > > It is very clear, and not only did I say where it's stated in the book, > I explicitly spelled it out... >> >> On Fri, Mar 26, 2010 at 12:29 AM, Chris Staub wrote: >>> On 03/25/2010 10:45 PM, John Stephens wrote: At this point, I believe all is OK. GCC is then unpacked in $LFS/gcc-build, mpfr-3.4.3 and gmp-5.0.0 are unpacked into $LFS/gcc-build/gcc-4.4.3 as mpfr and gmp respectively. GCC is then configured using: l...@carina:/mnt/lfs/gcc-build/gcc-4.4.3$ ../gcc-4.4.3/configure --target=$LFS_TGT --disable-nls --disable-shared --disable-multilib --disable-decimal-float --disable-threads --disable-libmudflap --disable-libssp --disable-libgomp --enable-languages=c ...snip... configure: creating ./config.status config.status: creating Makefile >>> >>> I see 3 problems here... >>> >>> 1. The LFS book does not say to create gcc-build before unpacking the >>> source tarball. Seems you need to take a close look at page 5.3, in >>> particular the very last "Note" there - it says that you are to unpack >>> the source tarball, cd into the created directory, *then* follow the >>> instructions on the page. >> >> Page 5.5.1 *DOES* say exactly that: >> "The GCC documentation recommends building GCC outside of the source >> directory in a dedicated build directory." >> >> Then in the next box: >> mkdir -v ../gcc-build >> cd ../gcc-build >> >> The line above the first box states that GCC requires GMP and MPFR. >> Now, if mpfr and gmp are unpacked in /mnt/lfs/sources, GCC has no >> access to them. The only way I could get GCC to do anything was to >> move both pkgs into /mnt/lfs/gcc-build/gcc-4.4.3/. But I am concerned >> about that. >> > I know exactly what the book says. I will repeat what I just said > above...if you look at the last note on page 5.3, it says that before > you follow any of the instructions on a package installation page, you > are to unpack that package's source and cd into the source directory. > So, for GCC, you unpack the gcc-4.4.3 tarball, cd gcc-4.4.3, *then* do > what the GCC page says. Again, this is in page 5.3 - however it is not > the *only* important piece of info on that page, so be sure to go there > and read it carefully - in fact, you might just want to keep page 5.3 > open and re-read it before every package. > >>> >>> 2. The gcc-build directory is not in $LFS/sources, so that means that >> >> Correct, but that is contrary to the instructions. > > Following what I just said above, if you are following the instructions > given on page 5.3, which means you are starting from the GCC source > directory (gcc-4.4.3) then gcc-build *will* be in $LFS/sources. > > Similarly, since the GCC page says to unpack GMP and MPFR *before* > gcc-build is created, that means you are in the GCC source dir when you > do so. > > So, to recap, unpack the GCC tarball, cd gcc-4.4.3, then do as stated on > the GCC installation page, starting with unpacking GMP and MPFR and > renaming their source dirs, then create and cd ../gcc-build. It is the > exact same procedure for every package - you unpack the tarball, cd into > the created source directory, then do as stated on the package's page, > followed by remo
Re: Bk6.6Ch6.9 glibc compile stopped with `__stack_chk_guard'
On 26 March 2010 04:06, wrote: > > My variance from book 6.6 consist of: > > /mnt/lfs/source - source files downloaded into this directory. > > /mnt/lfs/scratch - source files un-tar'ed and built in this > directory. the above seem reasonable > > script files - to reduce opps's while cutting and pasting > lines to execute i placed them in script > files in order to ensure correct transcription > from the book. I would then execute the > script and verify that it did what it was > suppose to. > Did you do anything to ensure that the scripts all stop on errors ? If you didn't, I guess you've had an error in an earlier stage. The references to __stack_chk_guard crop up over the years. Generally, the problem seems to be from a bad build environment, i.e. "something is wrong somewhere". ĸen -- After tragedy, and farce, "OMG poneys!" -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
On Thu, Mar 25, 2010 at 11:05 PM, Bruce Dubbs wrote: > > John Stephens wrote: > > > > checking for gawk... no > > checking for mawk... mawk > > Make sure you have all the prereqs as stated in section vii. Host System > Requirements, especially awk->gawk, yacc->bison, and sh->bash. > > -- Bruce Bruce, Thanks for finding that. I had gone thru that list (host sys req.) and I nothing popped out as being unusual. I missed that Ubuntu doesn't supply those routinely. My other distros did -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: grub2: error: fiel not found
On 26 March 2010 08:34, Simon Geard wrote: > On Thu, 2010-03-25 at 20:04 +, Ken Moffat wrote: >> No, this is grub2 - on x86, disks still start at zero, but dos partitions >> have their "real" number (1-4 for primary, 5+ for extended) so (hd0,15) >> is indeed sda15. > > Yeah... because having them inconsistent is less confusing than having > them both start at zero? > > You say *dos* partitions - does grub2 behave differently for other types > of partition table? > The syntax can differ. The one I noticed on my initial searches was for ppc macs using OpenFirmware and apple partitions. [ http://grub.enbug.org/TestingOnPowerPC ] linux (hd,7)/vmlinux root=/dev/hda9 For BSD partitions, letters are used, or slice numbers and letters. All of these variants seem to make the partition identifiers consistent with most other programs. ĸen -- After tragedy, and farce, "OMG poneys!" -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
On 03/26/2010 07:04 AM, John Stephens wrote: > Chris, > Thanks for the reply. I've spent a lot of time in chapters 4 and 5. > I'm finding the book (6.6) a little unclear in certain areas which is > probably what is throwing me off to some degree. > > For example, look at 5.5.1, first box. It says: > > tar -jxf ../mpfr-2.4.2.tar.bz2 > mv -v mpfr-2.4.2 mpfr > tar -jxf ../gmp-5.0.0.tar.bz2 > mv -v ../gmp-5.0.0 gmp > > What is the point of reference for "../pkg-name"? If I am in my > /mnt/lfs/sources dir, then the ../ takes me up to /mnt/lfs and the > source tarball is not there. > > I haven't found any reference as to where I should be sitting > (directory wise...) when I issue the commands. It is very clear, and not only did I say where it's stated in the book, I explicitly spelled it out... > > On Fri, Mar 26, 2010 at 12:29 AM, Chris Staub wrote: >> On 03/25/2010 10:45 PM, John Stephens wrote: >>> >>> At this point, I believe all is OK. GCC is then unpacked in >>> $LFS/gcc-build, mpfr-3.4.3 and gmp-5.0.0 are unpacked into >>> $LFS/gcc-build/gcc-4.4.3 as mpfr and gmp respectively. >>> >>> GCC is then configured using: >>> l...@carina:/mnt/lfs/gcc-build/gcc-4.4.3$ ../gcc-4.4.3/configure >>> --target=$LFS_TGT --disable-nls --disable-shared --disable-multilib >>> --disable-decimal-float --disable-threads --disable-libmudflap >>> --disable-libssp --disable-libgomp --enable-languages=c >>> ...snip... >>> configure: creating ./config.status >>> config.status: creating Makefile >> >> I see 3 problems here... >> >> 1. The LFS book does not say to create gcc-build before unpacking the >> source tarball. Seems you need to take a close look at page 5.3, in >> particular the very last "Note" there - it says that you are to unpack >> the source tarball, cd into the created directory, *then* follow the >> instructions on the page. > > Page 5.5.1 *DOES* say exactly that: > "The GCC documentation recommends building GCC outside of the source > directory in a dedicated build directory." > > Then in the next box: > mkdir -v ../gcc-build > cd ../gcc-build > > The line above the first box states that GCC requires GMP and MPFR. > Now, if mpfr and gmp are unpacked in /mnt/lfs/sources, GCC has no > access to them. The only way I could get GCC to do anything was to > move both pkgs into /mnt/lfs/gcc-build/gcc-4.4.3/. But I am concerned > about that. > I know exactly what the book says. I will repeat what I just said above...if you look at the last note on page 5.3, it says that before you follow any of the instructions on a package installation page, you are to unpack that package's source and cd into the source directory. So, for GCC, you unpack the gcc-4.4.3 tarball, cd gcc-4.4.3, *then* do what the GCC page says. Again, this is in page 5.3 - however it is not the *only* important piece of info on that page, so be sure to go there and read it carefully - in fact, you might just want to keep page 5.3 open and re-read it before every package. >> >> 2. The gcc-build directory is not in $LFS/sources, so that means that > > Correct, but that is contrary to the instructions. Following what I just said above, if you are following the instructions given on page 5.3, which means you are starting from the GCC source directory (gcc-4.4.3) then gcc-build *will* be in $LFS/sources. Similarly, since the GCC page says to unpack GMP and MPFR *before* gcc-build is created, that means you are in the GCC source dir when you do so. So, to recap, unpack the GCC tarball, cd gcc-4.4.3, then do as stated on the GCC installation page, starting with unpacking GMP and MPFR and renaming their source dirs, then create and cd ../gcc-build. It is the exact same procedure for every package - you unpack the tarball, cd into the created source directory, then do as stated on the package's page, followed by removing the source and (and build, if created) dirs (which is explained in another note on page 5.3). For the same reason, binutils-build should also be located in $LFS/sources. -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
Le vendredi 26 mars à 12:04, John Stephens a écrit : > > Chris, > Thanks for the reply. I've spent a lot of time in chapters 4 and 5. > I'm finding the book (6.6) a little unclear in certain areas which is > probably what is throwing me off to some degree. > > For example, look at 5.5.1, first box. It says: > > tar -jxf ../mpfr-2.4.2.tar.bz2 > mv -v mpfr-2.4.2 mpfr > tar -jxf ../gmp-5.0.0.tar.bz2 > mv -v ../gmp-5.0.0 gmp > > What is the point of reference for "../pkg-name"? If I am in my > /mnt/lfs/sources dir, then the ../ takes me up to /mnt/lfs and the > source tarball is not there. It's very clear, you didn't read the end of 5.3 !!! -- Ph. Delavalade -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Unable to compile GCC-4.4.3, Pass 1
Chris, Thanks for the reply. I've spent a lot of time in chapters 4 and 5. I'm finding the book (6.6) a little unclear in certain areas which is probably what is throwing me off to some degree. For example, look at 5.5.1, first box. It says: tar -jxf ../mpfr-2.4.2.tar.bz2 mv -v mpfr-2.4.2 mpfr tar -jxf ../gmp-5.0.0.tar.bz2 mv -v ../gmp-5.0.0 gmp What is the point of reference for "../pkg-name"? If I am in my /mnt/lfs/sources dir, then the ../ takes me up to /mnt/lfs and the source tarball is not there. I haven't found any reference as to where I should be sitting (directory wise...) when I issue the commands. On Fri, Mar 26, 2010 at 12:29 AM, Chris Staub wrote: > On 03/25/2010 10:45 PM, John Stephens wrote: >> Greetings, >> >> I am attempting to build my first LFS project, but can't see to get past >> the GCC, pass 1 compile. FWIW, Linux is not new to me. I have had an >> interest in working through a LFS project for several years, and now I >> actually have some time to do it. >> >> I suspect I have a configuration setting gone wrong, however I am unable >> to see what is going wrong. >> >> I have a Ubuntu 9.10 host (AMD Athlon CPU, 1G memory). I've defined an >> lsf user and have a separate 50G partition mounted at /mnt/lfs. All the >> suggested symlinks are in place. >> >> >> At this point, I believe all is OK. GCC is then unpacked in >> $LFS/gcc-build, mpfr-3.4.3 and gmp-5.0.0 are unpacked into >> $LFS/gcc-build/gcc-4.4.3 as mpfr and gmp respectively. >> >> GCC is then configured using: >> l...@carina:/mnt/lfs/gcc-build/gcc-4.4.3$ ../gcc-4.4.3/configure >> --target=$LFS_TGT --disable-nls --disable-shared --disable-multilib >> --disable-decimal-float --disable-threads --disable-libmudflap >> --disable-libssp --disable-libgomp --enable-languages=c >> ...snip... >> configure: creating ./config.status >> config.status: creating Makefile > > I see 3 problems here... > > 1. The LFS book does not say to create gcc-build before unpacking the > source tarball. Seems you need to take a close look at page 5.3, in > particular the very last "Note" there - it says that you are to unpack > the source tarball, cd into the created directory, *then* follow the > instructions on the page. Page 5.5.1 *DOES* say exactly that: "The GCC documentation recommends building GCC outside of the source directory in a dedicated build directory." Then in the next box: mkdir -v ../gcc-build cd ../gcc-build The line above the first box states that GCC requires GMP and MPFR. Now, if mpfr and gmp are unpacked in /mnt/lfs/sources, GCC has no access to them. The only way I could get GCC to do anything was to move both pkgs into /mnt/lfs/gcc-build/gcc-4.4.3/. But I am concerned about that. > > 2. The gcc-build directory is not in $LFS/sources, so that means that Correct, but that is contrary to the instructions. > either you have become root (or used sudo) to create it, or you have > chowned/chmodded $LFS, neither of which you should be doing. If you did > this because you were getting permissions errors attempting to create > gcc-build, see #1 above that will solve that issue. Also, do not use > root/sudo at any point in Chapter 5 until you are told to, and if you > did chown/chmod $LFS, change it back (owned by root, 755 permissions). I've tried so many different things trying to get this to work that I will go back and rebuild the environment again. I don't believe I have had any permission problems, but I will backtrack just to be sure again. > > 3. You forgot --prefix on that configure command. That would explain the > permission errors on make install. Funny thing... during my last attempt (which I used to get the messages for this posting) I did forget the --prefix. > > I don't recognize the particular error you are getting and don't know > whether it's related to any of the above issues, but the fact that you > did miss those is an indicator that you do need to read a bit more > closely. The compile errors you are getting are likely due to some other > detail you've overlooked. Remove the GCC source and build dirs and try > again, this time reading more slowly and carefully. If that still > doesn't work, rm -rf /tools/* and go back to the beginning of Chapter 5. I'll check the prereqs again and rebuild the environment (again :) I've gone back and forth over chapters 4 and 5 so often that I can almost quote them. But obviously, I am missing something either conceptually or I am mis-interpreting something. > >> >> If I try to do a "make install", I get the following: >> >> l...@carina:/mnt/lfs/gcc-build/gcc-4.4.3$ make install >> make[1]: Entering directory `/mnt/lfs/gcc-build/gcc-4.4.3' >> /bin/bash ./mkinstalldirs /usr/local /usr/local >> make[2]: Entering directory >> `/mnt/lfs/gcc-build/gcc-4.4.3/host-i686-pc-linux-gnu/fixincludes' >> rm -rf /usr/local/libexec/gcc/i686-lfs-linux-gnu/4.4.3/install-tools >> /bin/bash ../.././fixincludes/../mkinstalldirs >> /usr/local/libexec/gcc/i686-lfs-linux-gnu/
Re: grub2: error: fiel not found
On Thu, 2010-03-25 at 20:04 +, Ken Moffat wrote: > On 25 March 2010 20:00, Mike McCarty wrote: > > > > > A nit, but nits are important in this context, shouldn't that be > > /dev/sda16 ? Or did I miss something? > > > > No, this is grub2 - on x86, disks still start at zero, but dos partitions > have their "real" number (1-4 for primary, 5+ for extended) so (hd0,15) > is indeed sda15. Yeah... because having them inconsistent is less confusing than having them both start at zero? You say *dos* partitions - does grub2 behave differently for other types of partition table? Simon. signature.asc Description: This is a digitally signed message part -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page