Re: 2.4.17 kernel & pcmcia problem
Heather <[EMAIL PROTECTED]> wrote: > I don't think orinoco cards can be classed as either rare or crappy > - but their module has definitely changed names, and *that* was > dependent on the kernel rev. Apples ... oranges. I was *not* talking about orinoco cards. However, since you have brought the subject up. Orinoco cards, like most cards, now have two sets of drivers. In this case, they are named differently. The 2.4 kernel driver is named "orinoco_cs", the standalone is named "wvlan_cs". Note that either driver works with the 2.4 kernels. > ... not solely though, I think. If the external pcmcia-cs modules > and the kernel-internal modules (for new kernels) are also named > differently, then auto-includes on kernel rev along wouldn't > necessarily work either. H... No. That's not the way to handle things. With the 2.4 kernels, most users can choose which set of drivers they want to use. (The exception being those with hardware that is supported by only one set of drivers.) The latest pcmcia-cs packages checks for the presence of the standalone drivers for the currently running kernel. If they are found, then cardmgr is configured to use those. Otherwise, it is configured to use the 2.4 kernel drivers. That is my current solution. I welcome feedback from anyone explaining why it is not sufficient. Please, however, provide concrete examples of what is failing and when it fails. - Brian
Re: 2.4.17 kernel & pcmcia problem
> > > or laptops. Examples of these special cases can be found in > > > /usr/share/doc/pcmcia-cs/config.opts.gz. > > To which Heather <[EMAIL PROTECTED]> commented: > > > Is it me, or does "only a few cases will need hand hacking" sound like > > it *does not satisfy* "folks with any kernel-image-n.n.nn or even > > handbuilt kernels shouldn't fail" ? > > They won't fail. The "special cases" are folks with hardware that does > not like the settings used by the default cardmgr configuration. Let's > face it: some hardware is difficult to handle; some hardware is just > plain crappy. > > > Can we split these changed-cases off into an extra support file so it > > can be sought by kernel version, or something? Whatta mess. > > This is not dependent on kernel version. It depends on the hardware > being used. These tweaks cannot be easily handled by automatic means. > They require experimentation on the part of the user. Fortunately, > these cases are rare. > > - Brian I don't think orinoco cards can be classed as either rare or crappy - but their module has definitely changed names, and *that* was dependent on the kernel rev. ... not solely though, I think. If the external pcmcia-cs modules and the kernel-internal modules (for new kernels) are also named differently, then auto-includes on kernel rev along wouldn't necessarily work either. H... * Heather Stern * star@ many places...
Re: 2.4.17 kernel & pcmcia problem
Heather <[EMAIL PROTECTED]> wrote: > I don't think orinoco cards can be classed as either rare or crappy > - but their module has definitely changed names, and *that* was > dependent on the kernel rev. Apples ... oranges. I was *not* talking about orinoco cards. However, since you have brought the subject up. Orinoco cards, like most cards, now have two sets of drivers. In this case, they are named differently. The 2.4 kernel driver is named "orinoco_cs", the standalone is named "wvlan_cs". Note that either driver works with the 2.4 kernels. > ... not solely though, I think. If the external pcmcia-cs modules > and the kernel-internal modules (for new kernels) are also named > differently, then auto-includes on kernel rev along wouldn't > necessarily work either. H... No. That's not the way to handle things. With the 2.4 kernels, most users can choose which set of drivers they want to use. (The exception being those with hardware that is supported by only one set of drivers.) The latest pcmcia-cs packages checks for the presence of the standalone drivers for the currently running kernel. If they are found, then cardmgr is configured to use those. Otherwise, it is configured to use the 2.4 kernel drivers. That is my current solution. I welcome feedback from anyone explaining why it is not sufficient. Please, however, provide concrete examples of what is failing and when it fails. - Brian -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
> > > or laptops. Examples of these special cases can be found in > > > /usr/share/doc/pcmcia-cs/config.opts.gz. > > To which Heather <[EMAIL PROTECTED]> commented: > > > Is it me, or does "only a few cases will need hand hacking" sound like > > it *does not satisfy* "folks with any kernel-image-n.n.nn or even > > handbuilt kernels shouldn't fail" ? > > They won't fail. The "special cases" are folks with hardware that does > not like the settings used by the default cardmgr configuration. Let's > face it: some hardware is difficult to handle; some hardware is just > plain crappy. > > > Can we split these changed-cases off into an extra support file so it > > can be sought by kernel version, or something? Whatta mess. > > This is not dependent on kernel version. It depends on the hardware > being used. These tweaks cannot be easily handled by automatic means. > They require experimentation on the part of the user. Fortunately, > these cases are rare. > > - Brian I don't think orinoco cards can be classed as either rare or crappy - but their module has definitely changed names, and *that* was dependent on the kernel rev. ... not solely though, I think. If the external pcmcia-cs modules and the kernel-internal modules (for new kernels) are also named differently, then auto-includes on kernel rev along wouldn't necessarily work either. H... * Heather Stern * star@ many places... -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
Tom Allison <[EMAIL PROTECTED]> wrote: > I have managed to get this working with 3.1.31-7. However, my kernel > drivers are selected for a driver 'orinoco' while the pcmcia package > based drivers are wlan_cs. The wlan_cs drivers are the older drivers > for the D-Link DWL-650 and are being replaced by these orinoco > drivers. It's a real problem whenever I try to load in a new kernel > and have all these drivers conflict like this. > Using the pcmcia-source, I was able to get the card working, but only > as a wvlan_cs driver based card. > Using the kernel drivers and the upgraded pcmcia-cs utilities, I was > able to get back to the orinoco drivers, for the time being. As I have mentioned before, if you want to use the new drivers (the "orinoco" driver not the "wlan_cs" driver), then you do not need to install a pcmcia-modules-2.4.17 package. The pcmcia-cs package has been fixed to support these new drivers. Remove the pcmcia-modules package and the conflicts will go away. - Brian
Re: 2.4.17 kernel & pcmcia problem
Tom Allison <[EMAIL PROTECTED]> wrote: > I have managed to get this working with 3.1.31-7. However, my kernel > drivers are selected for a driver 'orinoco' while the pcmcia package > based drivers are wlan_cs. The wlan_cs drivers are the older drivers > for the D-Link DWL-650 and are being replaced by these orinoco > drivers. It's a real problem whenever I try to load in a new kernel > and have all these drivers conflict like this. > Using the pcmcia-source, I was able to get the card working, but only > as a wvlan_cs driver based card. > Using the kernel drivers and the upgraded pcmcia-cs utilities, I was > able to get back to the orinoco drivers, for the time being. As I have mentioned before, if you want to use the new drivers (the "orinoco" driver not the "wlan_cs" driver), then you do not need to install a pcmcia-modules-2.4.17 package. The pcmcia-cs package has been fixed to support these new drivers. Remove the pcmcia-modules package and the conflicts will go away. - Brian -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
Brian Mays wrote: Heather <[EMAIL PROTECTED]> wrote: [symlinks bug in kernel-image-* ellided] Well, since the dev gang are aware of this I'm sure kernel-image-2.4.18-* will be fixed when they happen. They should be. If not, complain to Herbert Xu. Brian Mays wrote: Try version 3.1.31-7. The only "hand hackery" that should be necessary is to deal with problem with specific special-case cards or laptops. Examples of these special cases can be found in /usr/share/doc/pcmcia-cs/config.opts.gz. I have managed to get this working with 3.1.31-7. However, my kernel drivers are selected for a driver 'orinoco' while the pcmcia package based drivers are wlan_cs. The wlan_cs drivers are the older drivers for the D-Link DWL-650 and are being replaced by these orinoco drivers. It's a real problem whenever I try to load in a new kernel and have all these drivers conflict like this. Using the pcmcia-source, I was able to get the card working, but only as a wvlan_cs driver based card. Using the kernel drivers and the upgraded pcmcia-cs utilities, I was able to get back to the orinoco drivers, for the time being. Of course, this will probably all break as I'm recompiling my kernel for sound support. It seems that I never did have the correct drivers in my kernel!
Re: 2.4.17 kernel & pcmcia problem
Brian Mays wrote: > Heather <[EMAIL PROTECTED]> wrote: > > >>[symlinks bug in kernel-image-* ellided] Well, since the dev gang are >>aware of this I'm sure kernel-image-2.4.18-* will be fixed when they >>happen. >> > > They should be. If not, complain to Herbert Xu. > > Brian Mays wrote: > > >>>Try version 3.1.31-7. The only "hand hackery" that should be >>>necessary is to deal with problem with specific special-case cards >>>or laptops. Examples of these special cases can be found in >>>/usr/share/doc/pcmcia-cs/config.opts.gz. >>> > I have managed to get this working with 3.1.31-7. However, my kernel drivers are selected for a driver 'orinoco' while the pcmcia package based drivers are wlan_cs. The wlan_cs drivers are the older drivers for the D-Link DWL-650 and are being replaced by these orinoco drivers. It's a real problem whenever I try to load in a new kernel and have all these drivers conflict like this. Using the pcmcia-source, I was able to get the card working, but only as a wvlan_cs driver based card. Using the kernel drivers and the upgraded pcmcia-cs utilities, I was able to get back to the orinoco drivers, for the time being. Of course, this will probably all break as I'm recompiling my kernel for sound support. It seems that I never did have the correct drivers in my kernel! -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
Heather <[EMAIL PROTECTED]> wrote: > [symlinks bug in kernel-image-* ellided] Well, since the dev gang are > aware of this I'm sure kernel-image-2.4.18-* will be fixed when they > happen. They should be. If not, complain to Herbert Xu. Brian Mays wrote: > > Try version 3.1.31-7. The only "hand hackery" that should be > > necessary is to deal with problem with specific special-case cards > > or laptops. Examples of these special cases can be found in > > /usr/share/doc/pcmcia-cs/config.opts.gz. To which Heather <[EMAIL PROTECTED]> commented: > Is it me, or does "only a few cases will need hand hacking" sound like > it *does not satisfy* "folks with any kernel-image-n.n.nn or even > handbuilt kernels shouldn't fail" ? They won't fail. The "special cases" are folks with hardware that does not like the settings used by the default cardmgr configuration. Let's face it: some hardware is difficult to handle; some hardware is just plain crappy. > Can we split these changed-cases off into an extra support file so it > can be sought by kernel version, or something? Whatta mess. This is not dependent on kernel version. It depends on the hardware being used. These tweaks cannot be easily handled by automatic means. They require experimentation on the part of the user. Fortunately, these cases are rare. - Brian
Re: 2.4.17 kernel & pcmcia problem
> Heather <[EMAIL PROTECTED]> replied: > > > When the names of modules that are used by some -cards- changed > > between 2.2 and 2.4, and (gasp) if they're different between David > > Hind's external modules or Linus' internal ones, then yes, it > > certainly -is- going to bother pcmcia-cs. The config file will > > announce incorrect mappings. > > Beep, bonk. :( > > This is a problem, but it is not the specific problem that Tom was talking > about. [symlinks bug in kernel-image-* ellided] Well, since the dev gang are aware of this I'm sure kernel-image-2.4.18-* will be fixed when they happen. > > ... Not that I'm afraid of a little text editing but it needs to be made > > AWARE and do the right thing without hand hackery, so folks' laptops > > will "just work" when they change Debian kernel kits or even (per > > common wisdom) build their own. Even if it's just a debconf saying, > > which of 3 config layouts should I use? > > Try version 3.1.31-7. The only "hand hackery" that should be necessary is > to deal with problem with specific special-case cards or laptops. > Examples of these special cases can be found in /usr/share/doc/pcmcia-cs/co > nfig.opts.gz. Is it me, or does "only a few cases will need hand hacking" sound like it *does not satisfy* "folks with any kernel-image-n.n.nn or even handbuilt kernels shouldn't fail" ? Can we split these changed-cases off into an extra support file so it can be sought by kernel version, or something? Whatta mess. * Heather Stern * star@ many places...
Re: 2.4.17 kernel & pcmcia problem
Heather <[EMAIL PROTECTED]> wrote: > [symlinks bug in kernel-image-* ellided] Well, since the dev gang are > aware of this I'm sure kernel-image-2.4.18-* will be fixed when they > happen. They should be. If not, complain to Herbert Xu. Brian Mays wrote: > > Try version 3.1.31-7. The only "hand hackery" that should be > > necessary is to deal with problem with specific special-case cards > > or laptops. Examples of these special cases can be found in > > /usr/share/doc/pcmcia-cs/config.opts.gz. To which Heather <[EMAIL PROTECTED]> commented: > Is it me, or does "only a few cases will need hand hacking" sound like > it *does not satisfy* "folks with any kernel-image-n.n.nn or even > handbuilt kernels shouldn't fail" ? They won't fail. The "special cases" are folks with hardware that does not like the settings used by the default cardmgr configuration. Let's face it: some hardware is difficult to handle; some hardware is just plain crappy. > Can we split these changed-cases off into an extra support file so it > can be sought by kernel version, or something? Whatta mess. This is not dependent on kernel version. It depends on the hardware being used. These tweaks cannot be easily handled by automatic means. They require experimentation on the part of the user. Fortunately, these cases are rare. - Brian -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
> Heather <[EMAIL PROTECTED]> replied: > > > When the names of modules that are used by some -cards- changed > > between 2.2 and 2.4, and (gasp) if they're different between David > > Hind's external modules or Linus' internal ones, then yes, it > > certainly -is- going to bother pcmcia-cs. The config file will > > announce incorrect mappings. > > Beep, bonk. :( > > This is a problem, but it is not the specific problem that Tom was talking > about. [symlinks bug in kernel-image-* ellided] Well, since the dev gang are aware of this I'm sure kernel-image-2.4.18-* will be fixed when they happen. > > ... Not that I'm afraid of a little text editing but it needs to be made > > AWARE and do the right thing without hand hackery, so folks' laptops > > will "just work" when they change Debian kernel kits or even (per > > common wisdom) build their own. Even if it's just a debconf saying, > > which of 3 config layouts should I use? > > Try version 3.1.31-7. The only "hand hackery" that should be necessary is > to deal with problem with specific special-case cards or laptops. > Examples of these special cases can be found in /usr/share/doc/pcmcia-cs/co > nfig.opts.gz. Is it me, or does "only a few cases will need hand hacking" sound like it *does not satisfy* "folks with any kernel-image-n.n.nn or even handbuilt kernels shouldn't fail" ? Can we split these changed-cases off into an extra support file so it can be sought by kernel version, or something? Whatta mess. * Heather Stern * star@ many places... -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
Brian Mays wrote: > > I just checked to refresh my memory on the problem that you are > > discussing. If you are using the 2.4 kernel drivers, then you do > > not need to install (or even build) a pcmcia-modules-2.4.18 package. > > This is the package that had the conflicts with the kernel-image > > package. Since you do not need to install it, you should not > > encounter this problem again. > > > > The contents of the pcmcia-cs package should not have conflicted > > with anything. Therefore, this is not a problem with pcmcia-cs. Heather <[EMAIL PROTECTED]> replied: > When the names of modules that are used by some -cards- changed > between 2.2 and 2.4, and (gasp) if they're different between David > Hind's external modules or Linus' internal ones, then yes, it > certainly -is- going to bother pcmcia-cs. The config file will > announce incorrect mappings. > Beep, bonk. :( This is a problem, but it is not the specific problem that Tom was talking about. He was referring to the conflict between the symlinks in the /lib/modules//pcmcia directory of the current kernel-image-* packages and the files in the same directory of the pcmcia-modules-* package. This is the correct location for the standalone drivers, and the symlinks are not necessary. Therefore, this is a problem with the kernel-image-* packages, and it has nothing to do with pcmcia-cs. > ... Not that I'm afraid of a little text editing but it needs to be made > AWARE and do the right thing without hand hackery, so folks' laptops > will "just work" when they change Debian kernel kits or even (per > common wisdom) build their own. Even if it's just a debconf saying, > which of 3 config layouts should I use? Try version 3.1.31-7. The only "hand hackery" that should be necessary is to deal with problem with specific special-case cards or laptops. Examples of these special cases can be found in /usr/share/doc/pcmcia-cs/co nfig.opts.gz. - Brian
Re: 2.4.17 kernel & pcmcia problem
> I just checked to refresh my memory on the problem that you are > discussing. If you are using the 2.4 kernel drivers, then you do not need > to install (or even build) a pcmcia-modules-2.4.18 package. This is the > package that had the conflicts with the kernel-image package. Since you > do not need to install it, you should not encounter this problem again. > > The contents of the pcmcia-cs package should not have conflicted with > anything. Therefore, this is not a problem with pcmcia-cs. When the names of modules that are used by some -cards- changed between 2.2 and 2.4, and (gasp) if they're different between David Hind's external modules or Linus' internal ones, then yes, it certainly -is- going to bother pcmcia-cs. The config file will announce incorrect mappings. Beep, bonk. :( My most personal experience with this was awhile ago, when there was some question about whether the ATAPI card module was named ide-cs or ide_cs. Of course I'm one of few people who uses pcmcia or cardbus ATAPI rather than just using USB for that stuff. But it's the best behaved cdrw I've got access to (with the exception of cd-bizcards, they're misbalanced) so it's important to *me* anyway... Not that I'm afraid of a little text editing but it needs to be made AWARE and do the right thing without hand hackery, so folks' laptops will "just work" when they change Debian kernel kits or even (per common wisdom) build their own. Even if it's just a debconf saying, which of 3 config layouts should I use? * Heather Stern * star@ many places...
Re: 2.4.17 kernel & pcmcia problem
Brian Mays wrote: > > I just checked to refresh my memory on the problem that you are > > discussing. If you are using the 2.4 kernel drivers, then you do > > not need to install (or even build) a pcmcia-modules-2.4.18 package. > > This is the package that had the conflicts with the kernel-image > > package. Since you do not need to install it, you should not > > encounter this problem again. > > > > The contents of the pcmcia-cs package should not have conflicted > > with anything. Therefore, this is not a problem with pcmcia-cs. Heather <[EMAIL PROTECTED]> replied: > When the names of modules that are used by some -cards- changed > between 2.2 and 2.4, and (gasp) if they're different between David > Hind's external modules or Linus' internal ones, then yes, it > certainly -is- going to bother pcmcia-cs. The config file will > announce incorrect mappings. > Beep, bonk. :( This is a problem, but it is not the specific problem that Tom was talking about. He was referring to the conflict between the symlinks in the /lib/modules//pcmcia directory of the current kernel-image-* packages and the files in the same directory of the pcmcia-modules-* package. This is the correct location for the standalone drivers, and the symlinks are not necessary. Therefore, this is a problem with the kernel-image-* packages, and it has nothing to do with pcmcia-cs. > ... Not that I'm afraid of a little text editing but it needs to be made > AWARE and do the right thing without hand hackery, so folks' laptops > will "just work" when they change Debian kernel kits or even (per > common wisdom) build their own. Even if it's just a debconf saying, > which of 3 config layouts should I use? Try version 3.1.31-7. The only "hand hackery" that should be necessary is to deal with problem with specific special-case cards or laptops. Examples of these special cases can be found in /usr/share/doc/pcmcia-cs/co nfig.opts.gz. - Brian -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
> I just checked to refresh my memory on the problem that you are > discussing. If you are using the 2.4 kernel drivers, then you do not need > to install (or even build) a pcmcia-modules-2.4.18 package. This is the > package that had the conflicts with the kernel-image package. Since you > do not need to install it, you should not encounter this problem again. > > The contents of the pcmcia-cs package should not have conflicted with > anything. Therefore, this is not a problem with pcmcia-cs. When the names of modules that are used by some -cards- changed between 2.2 and 2.4, and (gasp) if they're different between David Hind's external modules or Linus' internal ones, then yes, it certainly -is- going to bother pcmcia-cs. The config file will announce incorrect mappings. Beep, bonk. :( My most personal experience with this was awhile ago, when there was some question about whether the ATAPI card module was named ide-cs or ide_cs. Of course I'm one of few people who uses pcmcia or cardbus ATAPI rather than just using USB for that stuff. But it's the best behaved cdrw I've got access to (with the exception of cd-bizcards, they're misbalanced) so it's important to *me* anyway... Not that I'm afraid of a little text editing but it needs to be made AWARE and do the right thing without hand hackery, so folks' laptops will "just work" when they change Debian kernel kits or even (per common wisdom) build their own. Even if it's just a debconf saying, which of 3 config layouts should I use? * Heather Stern * star@ many places... -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
Tom Allison <[EMAIL PROTECTED]> wrote: > I upgrade pcmcia-cs to pull in the newer applications (cardctl > v3.1.31) and in that process ended up with this problem. > > In the past 2.4.xx kernels I have been using the kernel supplied > drivers for my network cards. Is it safe to assume that the problem I > have experienced was limited to the fact that I also upgraded my > pcmcia-cs binaries. And, if I were to make a new kernel based on > 2.4.18 that I should not expect this problem again since I will be > using the kernel based pcmcia driver libraries? I just checked to refresh my memory on the problem that you are discussing. If you are using the 2.4 kernel drivers, then you do not need to install (or even build) a pcmcia-modules-2.4.18 package. This is the package that had the conflicts with the kernel-image package. Since you do not need to install it, you should not encounter this problem again. The contents of the pcmcia-cs package should not have conflicted with anything. Therefore, this is not a problem with pcmcia-cs. Those users who prefer the standalone drivers in the pcmcia-modules-* packages will continue to see these conflicts until the kernel packages are fixed to get rid of the symbolic links in the pcmcia subdirectory. If these users build their own custom kernels, they should *disable* the building of the 2.4 kernel drivers when configuring their kernels. This is done by saying "no" to PCMCIA support (i.e., the question about CONFIG_PCMCIA). This question is about the kernel-supplied drivers, and has nothing to do with the standalone drivers provided by the pcmcia-cs source. - Brian
Re: 2.4.17 kernel & pcmcia problem
Tom Allison <[EMAIL PROTECTED]> wrote: > I upgrade pcmcia-cs to pull in the newer applications (cardctl > v3.1.31) and in that process ended up with this problem. > > In the past 2.4.xx kernels I have been using the kernel supplied > drivers for my network cards. Is it safe to assume that the problem I > have experienced was limited to the fact that I also upgraded my > pcmcia-cs binaries. And, if I were to make a new kernel based on > 2.4.18 that I should not expect this problem again since I will be > using the kernel based pcmcia driver libraries? I just checked to refresh my memory on the problem that you are discussing. If you are using the 2.4 kernel drivers, then you do not need to install (or even build) a pcmcia-modules-2.4.18 package. This is the package that had the conflicts with the kernel-image package. Since you do not need to install it, you should not encounter this problem again. The contents of the pcmcia-cs package should not have conflicted with anything. Therefore, this is not a problem with pcmcia-cs. Those users who prefer the standalone drivers in the pcmcia-modules-* packages will continue to see these conflicts until the kernel packages are fixed to get rid of the symbolic links in the pcmcia subdirectory. If these users build their own custom kernels, they should *disable* the building of the 2.4 kernel drivers when configuring their kernels. This is done by saying "no" to PCMCIA support (i.e., the question about CONFIG_PCMCIA). This question is about the kernel-supplied drivers, and has nothing to do with the standalone drivers provided by the pcmcia-cs source. - Brian -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
Brian Mays wrote: Tom Allison <[EMAIL PROTECTED]> asked: QUESTION: If I were to use the kernel from ftp.kernel.org and then run a make-kpkg, would I see this same problem? It depends. If you are using the 2.4 kernel drivers (i.e., you have turned on the CONFIG_PCMCIA option) then you would not experience these problems. The problems with 3.1.31-6 occur only with those trying to use the new pcmcia-cs package with old versions of the pcmcia-modules-* packages or are trying to use the standalone modules without packaging them in a Debian package first. I upgrade pcmcia-cs to pull in the newer applications (cardctl v3.1.31) and in that process ended up with this problem. In the past 2.4.xx kernels I have been using the kernel supplied drivers for my network cards. Is it safe to assume that the problem I have experienced was limited to the fact that I also upgraded my pcmcia-cs binaries. And, if I were to make a new kernel based on 2.4.18 that I should not expect this problem again since I will be using the kernel based pcmcia driver libraries? Is that about the 'gist' of it?
Re: 2.4.17 kernel & pcmcia problem
Brian Mays wrote: > Tom Allison <[EMAIL PROTECTED]> asked: > > >>QUESTION: If I were to use the kernel from ftp.kernel.org and then >>run a make-kpkg, would I see this same problem? >> > > It depends. If you are using the 2.4 kernel drivers (i.e., you have > turned on the CONFIG_PCMCIA option) then you would not experience these > problems. The problems with 3.1.31-6 occur only with those trying to > use the new pcmcia-cs package with old versions of the pcmcia-modules-* > packages or are trying to use the standalone modules without packaging > them in a Debian package first. > > I upgrade pcmcia-cs to pull in the newer applications (cardctl v3.1.31) and in that process ended up with this problem. In the past 2.4.xx kernels I have been using the kernel supplied drivers for my network cards. Is it safe to assume that the problem I have experienced was limited to the fact that I also upgraded my pcmcia-cs binaries. And, if I were to make a new kernel based on 2.4.18 that I should not expect this problem again since I will be using the kernel based pcmcia driver libraries? Is that about the 'gist' of it? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
Tom Allison <[EMAIL PROTECTED]> asked: > QUESTION: If I were to use the kernel from ftp.kernel.org and then > run a make-kpkg, would I see this same problem? It depends. If you are using the 2.4 kernel drivers (i.e., you have turned on the CONFIG_PCMCIA option) then you would not experience these problems. The problems with 3.1.31-6 occur only with those trying to use the new pcmcia-cs package with old versions of the pcmcia-modules-* packages or are trying to use the standalone modules without packaging them in a Debian package first. > Also, I didn't see 3.1.36-7 in testing, yet. So I have gone ahead and > tried your suggestion of copying the config file over. It is tagged with an urgency of "high", but that still means that it will take several days to migrate into woody. The package needs to be built for the other architectures, not just for i386, before it is a candidate for the testing distribution. The package is available now in the unstable distribution. - Brian
Re: 2.4.17 kernel & pcmcia problem
On Sun, Feb 17, 2002 at 02:17:35PM -0500, Michael Stone wrote: > On Sun, Feb 17, 2002 at 08:23:26AM -0500, Pann McCuaig wrote: > > I had a similar problem. I turned off all PCMCIA support in the kernel. > > I don't know why it's there in the first place (seems to be new with > > 2.4.x). > > Well, I've got some machines that only work with the kernel pcmcia > drivers and some that only work with the external pcmcia drivers. > YMMV... > The way I understood it, broadly speaking you'd want the kernel drivers to handle the modern 32-bit cardbus cards, but you'd be better off using the standalone drivers for the older 16-bit pcmcia cards. Is this valid, or an over-simplification? Drew
Re: 2.4.17 kernel & pcmcia problem
Tom Allison <[EMAIL PROTECTED]> asked: > QUESTION: If I were to use the kernel from ftp.kernel.org and then > run a make-kpkg, would I see this same problem? It depends. If you are using the 2.4 kernel drivers (i.e., you have turned on the CONFIG_PCMCIA option) then you would not experience these problems. The problems with 3.1.31-6 occur only with those trying to use the new pcmcia-cs package with old versions of the pcmcia-modules-* packages or are trying to use the standalone modules without packaging them in a Debian package first. > Also, I didn't see 3.1.36-7 in testing, yet. So I have gone ahead and > tried your suggestion of copying the config file over. It is tagged with an urgency of "high", but that still means that it will take several days to migrate into woody. The package needs to be built for the other architectures, not just for i386, before it is a candidate for the testing distribution. The package is available now in the unstable distribution. - Brian -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
Brian Mays wrote: Tom Allison <[EMAIL PROTECTED]> wrote: But I built my own pcmcia-modules package and kernel-image package from the 2.4.17 kernel-source and 3.1.36 pcmcia-source. You mean that these are still building bad links? Not the pcmcia source. The kernel source is building the bad links, and there has not been a new release of this package since this problem was discovered. Therefore, there has not been an opportunity to fix the problem with the kernel packages. I have tried using the --force-overwrite option when installating the pcmcia-modules package and it still does not work. I am able to get past the problem with the pcmcia-core problem. But the specific drivers I am using (orinoco, orinoco_cs) are still failing to modprobe successfully. The temporary fix (until version 3.1.36-7 can be installed) is to copy the etc/config file from the pcmcia-cs source (i.e., /usr/src/modules/pcmcia/etc/config in the pcmcia-source package) to /etc/pcmcia/config. That should fix your problem for now. Your current config (assuming that you are running version 3.1.36-6 of pcmcia-cs) is set up to use the kernel drivers, not the standalone drivers. (This has been changed in 3.1.36-7.) Since you are building the pcmcia modules from the pcmcia-cs source, you want to use the standalone drivers. - Brian QUESTION: If I were to use the kernel from ftp.kernel.org and then run a make-kpkg, would I see this same problem? Also, I didn't see 3.1.36-7 in testing, yet. So I have gone ahead and tried your suggestion of copying the config file over. It works. What I was seeing is that I would get a successful load and the characteristic two Beeps. But then my log will fill up with Tx Errors as I tried to ping my gateway. Thank you very much!
Re: 2.4.17 kernel & pcmcia problem
On Sun, Feb 17, 2002 at 02:17:35PM -0500, Michael Stone wrote: > On Sun, Feb 17, 2002 at 08:23:26AM -0500, Pann McCuaig wrote: > > I had a similar problem. I turned off all PCMCIA support in the kernel. > > I don't know why it's there in the first place (seems to be new with > > 2.4.x). > > Well, I've got some machines that only work with the kernel pcmcia > drivers and some that only work with the external pcmcia drivers. > YMMV... > The way I understood it, broadly speaking you'd want the kernel drivers to handle the modern 32-bit cardbus cards, but you'd be better off using the standalone drivers for the older 16-bit pcmcia cards. Is this valid, or an over-simplification? Drew -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
Brian Mays wrote: > Tom Allison <[EMAIL PROTECTED]> wrote: > > >>But I built my own pcmcia-modules package and kernel-image package >>from the 2.4.17 kernel-source and 3.1.36 pcmcia-source. You mean that >>these are still building bad links? >> > > Not the pcmcia source. The kernel source is building the bad links, > and there has not been a new release of this package since this problem > was discovered. Therefore, there has not been an opportunity to fix the > problem with the kernel packages. > > >>I have tried using the --force-overwrite option when installating >>the pcmcia-modules package and it still does not work. I am able to >>get past the problem with the pcmcia-core problem. But the specific >>drivers I am using (orinoco, orinoco_cs) are still failing to modprobe >>successfully. >> > > The temporary fix (until version 3.1.36-7 can be installed) is > to copy the etc/config file from the pcmcia-cs source (i.e., > /usr/src/modules/pcmcia/etc/config in the pcmcia-source package) to > /etc/pcmcia/config. That should fix your problem for now. > > Your current config (assuming that you are running version 3.1.36-6 > of pcmcia-cs) is set up to use the kernel drivers, not the standalone > drivers. (This has been changed in 3.1.36-7.) Since you are building > the pcmcia modules from the pcmcia-cs source, you want to use the > standalone drivers. > > - Brian > > QUESTION: If I were to use the kernel from ftp.kernel.org and then run a make-kpkg, would I see this same problem? Also, I didn't see 3.1.36-7 in testing, yet. So I have gone ahead and tried your suggestion of copying the config file over. It works. What I was seeing is that I would get a successful load and the characteristic two Beeps. But then my log will fill up with Tx Errors as I tried to ping my gateway. Thank you very much! -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
On Sun, Feb 17, 2002 at 08:23:26AM -0500, Pann McCuaig wrote: > I had a similar problem. I turned off all PCMCIA support in the kernel. > I don't know why it's there in the first place (seems to be new with > 2.4.x). Well, I've got some machines that only work with the kernel pcmcia drivers and some that only work with the external pcmcia drivers. YMMV... -- Mike Stone pgp61wDnIJ53H.pgp Description: PGP signature
Re: 2.4.17 kernel & pcmcia problem
Tom Allison <[EMAIL PROTECTED]> wrote: > But I built my own pcmcia-modules package and kernel-image package > from the 2.4.17 kernel-source and 3.1.36 pcmcia-source. You mean that > these are still building bad links? Not the pcmcia source. The kernel source is building the bad links, and there has not been a new release of this package since this problem was discovered. Therefore, there has not been an opportunity to fix the problem with the kernel packages. > I have tried using the --force-overwrite option when installating > the pcmcia-modules package and it still does not work. I am able to > get past the problem with the pcmcia-core problem. But the specific > drivers I am using (orinoco, orinoco_cs) are still failing to modprobe > successfully. The temporary fix (until version 3.1.36-7 can be installed) is to copy the etc/config file from the pcmcia-cs source (i.e., /usr/src/modules/pcmcia/etc/config in the pcmcia-source package) to /etc/pcmcia/config. That should fix your problem for now. Your current config (assuming that you are running version 3.1.36-6 of pcmcia-cs) is set up to use the kernel drivers, not the standalone drivers. (This has been changed in 3.1.36-7.) Since you are building the pcmcia modules from the pcmcia-cs source, you want to use the standalone drivers. - Brian
Re: 2.4.17 kernel & pcmcia problem
On Sun, Feb 17, 2002 at 08:23:26AM -0500, Pann McCuaig wrote: > I had a similar problem. I turned off all PCMCIA support in the kernel. > I don't know why it's there in the first place (seems to be new with > 2.4.x). Well, I've got some machines that only work with the kernel pcmcia drivers and some that only work with the external pcmcia drivers. YMMV... -- Mike Stone msg06757/pgp0.pgp Description: PGP signature
Re: 2.4.17 kernel & pcmcia problem
Brian Mays wrote: [EMAIL PROTECTED] (Tom Allison) wrote: I have built a kernel and modules images and have tried to install them both. I have to do a dpkp -i --force-overwrite in order to get them to install. The error is related to a duplicate of pcmcia_core modules. It's not a duplicate of the modules. Instead, there are two sets of drivers, the drivers supplied by the 2.4 kernel and the drivers provided by the pcmcia-cs source. The kernel-image package now, incorrectly, includes some symlinks from the /lib/modules//pcmcia directory to the kernel modules located elsewhere in the /lib/modules/ directory tree. Since these symlinks are located in the location where the standalone drivers have always resided, a conflict occurs and "--force-overwrite" is necessary. See Bug#128662 for more information. - Brian But I built my own pcmcia-modules package and kernel-image package from the 2.4.17 kernel-source and 3.1.36 pcmcia-source. You mean that these are still building bad links? I have tried using the --force-overwrite option when installating the pcmcia-modules package and it still does not work. I am able to get past the problem with the pcmcia-core problem. But the specific drivers I am using (orinoco, orinoco_cs) are still failing to modprobe successfully. Now what I am seeing is a versioning problem between these drivers in the two pacages.
Re: 2.4.17 kernel & pcmcia problem
Pann McCuaig wrote: > I had a similar problem. I turned off all PCMCIA support in the > kernel. I don't know why it's there in the first place (seems to be > new with 2.4.x). The 2.4 kernels have introduced their own PCMCIA drivers. Now there are two sets of drivers, the 2.4 kernel drivers, and the standalone drivers (from the pcmcia-cs source). I am currently in the process of trying to get one pcmcia-cs package to support both sets. [EMAIL PROTECTED] (Tom Allison) wrote: > Yeah, but it's been working for months and it seems to me that the > pcmcia drivers are going into the kernel rather than coming out. Personally, I still recommend the standalone drivers for now, since they are more mature. New development is almost completely in the 2.4 kernel drivers, however, so they are the way of the future. - Brian
Re: 2.4.17 kernel & pcmcia problem
[EMAIL PROTECTED] (Tom Allison) wrote: > I have built a kernel and modules images and have tried to install > them both. I have to do a dpkp -i --force-overwrite in order to get > them to install. The error is related to a duplicate of pcmcia_core > modules. It's not a duplicate of the modules. Instead, there are two sets of drivers, the drivers supplied by the 2.4 kernel and the drivers provided by the pcmcia-cs source. The kernel-image package now, incorrectly, includes some symlinks from the /lib/modules//pcmcia directory to the kernel modules located elsewhere in the /lib/modules/ directory tree. Since these symlinks are located in the location where the standalone drivers have always resided, a conflict occurs and "--force-overwrite" is necessary. See Bug#128662 for more information. - Brian
Re: 2.4.17 kernel & pcmcia problem
Pann McCuaig wrote: On Sat, Feb 16, 2002 at 23:13, Tom Allison wrote: Has anyone else noticed that the kernel-source for 2.4.17 and the pcmcia-source for 3.1.31 are not playing well with each other? I have found that the 2.4.17 orinoco driver (driver/net/wireless/) is version 08a and the pcmcia 3.1.31 orinoco driver is at version 08 (from /usr/src/modules/pcmcia-cs...). I had a similar problem. I turned off all PCMCIA support in the kernel. I don't know why it's there in the first place (seems to be new with 2.4.x). These little snippets are from my woody partition (I'm typing this on my potato partition). Yeah, but it's been working for months and it seems to me that the pcmcia drivers are going into the kernel rather than coming out.
Re: 2.4.17 kernel & pcmcia problem
Tom Allison <[EMAIL PROTECTED]> wrote: > But I built my own pcmcia-modules package and kernel-image package > from the 2.4.17 kernel-source and 3.1.36 pcmcia-source. You mean that > these are still building bad links? Not the pcmcia source. The kernel source is building the bad links, and there has not been a new release of this package since this problem was discovered. Therefore, there has not been an opportunity to fix the problem with the kernel packages. > I have tried using the --force-overwrite option when installating > the pcmcia-modules package and it still does not work. I am able to > get past the problem with the pcmcia-core problem. But the specific > drivers I am using (orinoco, orinoco_cs) are still failing to modprobe > successfully. The temporary fix (until version 3.1.36-7 can be installed) is to copy the etc/config file from the pcmcia-cs source (i.e., /usr/src/modules/pcmcia/etc/config in the pcmcia-source package) to /etc/pcmcia/config. That should fix your problem for now. Your current config (assuming that you are running version 3.1.36-6 of pcmcia-cs) is set up to use the kernel drivers, not the standalone drivers. (This has been changed in 3.1.36-7.) Since you are building the pcmcia modules from the pcmcia-cs source, you want to use the standalone drivers. - Brian -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
Brian Mays wrote: > [EMAIL PROTECTED] (Tom Allison) wrote: > > >>I have built a kernel and modules images and have tried to install >>them both. I have to do a dpkp -i --force-overwrite in order to get >>them to install. The error is related to a duplicate of pcmcia_core >>modules. >> > > It's not a duplicate of the modules. Instead, there are two sets of > drivers, the drivers supplied by the 2.4 kernel and the drivers provided > by the pcmcia-cs source. The kernel-image package now, incorrectly, > includes some symlinks from the /lib/modules//pcmcia directory to the > kernel modules located elsewhere in the /lib/modules/ directory tree. > Since these symlinks are located in the location where the standalone > drivers have always resided, a conflict occurs and "--force-overwrite" > is necessary. > > See Bug#128662 for more information. > > - Brian > > > But I built my own pcmcia-modules package and kernel-image package from the 2.4.17 kernel-source and 3.1.36 pcmcia-source. You mean that these are still building bad links? I have tried using the --force-overwrite option when installating the pcmcia-modules package and it still does not work. I am able to get past the problem with the pcmcia-core problem. But the specific drivers I am using (orinoco, orinoco_cs) are still failing to modprobe successfully. Now what I am seeing is a versioning problem between these drivers in the two pacages. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
Pann McCuaig wrote: > I had a similar problem. I turned off all PCMCIA support in the > kernel. I don't know why it's there in the first place (seems to be > new with 2.4.x). The 2.4 kernels have introduced their own PCMCIA drivers. Now there are two sets of drivers, the 2.4 kernel drivers, and the standalone drivers (from the pcmcia-cs source). I am currently in the process of trying to get one pcmcia-cs package to support both sets. [EMAIL PROTECTED] (Tom Allison) wrote: > Yeah, but it's been working for months and it seems to me that the > pcmcia drivers are going into the kernel rather than coming out. Personally, I still recommend the standalone drivers for now, since they are more mature. New development is almost completely in the 2.4 kernel drivers, however, so they are the way of the future. - Brian -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
[EMAIL PROTECTED] (Tom Allison) wrote: > I have built a kernel and modules images and have tried to install > them both. I have to do a dpkp -i --force-overwrite in order to get > them to install. The error is related to a duplicate of pcmcia_core > modules. It's not a duplicate of the modules. Instead, there are two sets of drivers, the drivers supplied by the 2.4 kernel and the drivers provided by the pcmcia-cs source. The kernel-image package now, incorrectly, includes some symlinks from the /lib/modules//pcmcia directory to the kernel modules located elsewhere in the /lib/modules/ directory tree. Since these symlinks are located in the location where the standalone drivers have always resided, a conflict occurs and "--force-overwrite" is necessary. See Bug#128662 for more information. - Brian -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
On Sat, Feb 16, 2002 at 23:13, Tom Allison wrote: > Has anyone else noticed that the kernel-source for 2.4.17 and the > pcmcia-source for 3.1.31 are not playing well with each other? > > I have found that the 2.4.17 orinoco driver (driver/net/wireless/) is > version 08a and the pcmcia 3.1.31 orinoco driver is at version 08 > (from /usr/src/modules/pcmcia-cs...). I had a similar problem. I turned off all PCMCIA support in the kernel. I don't know why it's there in the first place (seems to be new with 2.4.x). These little snippets are from my woody partition (I'm typing this on my potato partition). potato:~# grep -i pcmcia /mnt/usr/src/kernel-source-2.4.17/.config # PCMCIA/CardBus support # CONFIG_PCMCIA is not set # CONFIG_PHONE_IXJ_PCMCIA is not set potato:~# ls -lR /mnt/usr/src/modules/pcmcia-cs/ | grep orinoco -rw-rw-r--1 root src 99701 Oct 19 23:53 orinoco.c -rw-rw-r--1 root src 4107 Oct 9 23:01 orinoco.h -rw-rw-r--1 root src 23503 Oct 19 23:53 orinoco_cs.c potato:~# ls -lR /mnt/usr/src/kernel-source-2.4.17/ | grep orinoco -rw-r--r--1 root root 100085 Oct 9 18:13 orinoco.c -rw-r--r--1 root root 4145 Oct 9 18:13 orinoco.h -rw-r--r--1 root root23327 Oct 9 18:13 orinoco_cs.c -rw-r--r--1 root root10575 Dec 21 12:41 orinoco_plx.c Luck, Pann -- geek by nature, Linux by choice L I N U X .~. The Choice /V\ http://www.ourmanpann.com/linux/ of a GNU /( )\ Generation ^^-^^
Re: 2.4.17 kernel & pcmcia problem
Pann McCuaig wrote: > On Sat, Feb 16, 2002 at 23:13, Tom Allison wrote: > >>Has anyone else noticed that the kernel-source for 2.4.17 and the >>pcmcia-source for 3.1.31 are not playing well with each other? >> >>I have found that the 2.4.17 orinoco driver (driver/net/wireless/) is >>version 08a and the pcmcia 3.1.31 orinoco driver is at version 08 >>(from /usr/src/modules/pcmcia-cs...). >> > > I had a similar problem. I turned off all PCMCIA support in the kernel. > I don't know why it's there in the first place (seems to be new with > 2.4.x). > > These little snippets are from my woody partition (I'm typing this on my > potato partition). Yeah, but it's been working for months and it seems to me that the pcmcia drivers are going into the kernel rather than coming out. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.17 kernel & pcmcia problem
On Sat, Feb 16, 2002 at 23:13, Tom Allison wrote: > Has anyone else noticed that the kernel-source for 2.4.17 and the > pcmcia-source for 3.1.31 are not playing well with each other? > > I have found that the 2.4.17 orinoco driver (driver/net/wireless/) is > version 08a and the pcmcia 3.1.31 orinoco driver is at version 08 > (from /usr/src/modules/pcmcia-cs...). I had a similar problem. I turned off all PCMCIA support in the kernel. I don't know why it's there in the first place (seems to be new with 2.4.x). These little snippets are from my woody partition (I'm typing this on my potato partition). potato:~# grep -i pcmcia /mnt/usr/src/kernel-source-2.4.17/.config # PCMCIA/CardBus support # CONFIG_PCMCIA is not set # CONFIG_PHONE_IXJ_PCMCIA is not set potato:~# ls -lR /mnt/usr/src/modules/pcmcia-cs/ | grep orinoco -rw-rw-r--1 root src 99701 Oct 19 23:53 orinoco.c -rw-rw-r--1 root src 4107 Oct 9 23:01 orinoco.h -rw-rw-r--1 root src 23503 Oct 19 23:53 orinoco_cs.c potato:~# ls -lR /mnt/usr/src/kernel-source-2.4.17/ | grep orinoco -rw-r--r--1 root root 100085 Oct 9 18:13 orinoco.c -rw-r--r--1 root root 4145 Oct 9 18:13 orinoco.h -rw-r--r--1 root root23327 Oct 9 18:13 orinoco_cs.c -rw-r--r--1 root root10575 Dec 21 12:41 orinoco_plx.c Luck, Pann -- geek by nature, Linux by choice L I N U X .~. The Choice /V\ http://www.ourmanpann.com/linux/ of a GNU /( )\ Generation ^^-^^ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
