[gentoo-user] Re: kernel 2.6.25-r6 oddities; is this kernel really ready for stable?
Kevin O'Gorman wrote: I run gentoo x86 stable, so that I usually avoid this sort of thing. This kernel, however, looks balky to me, because it's reporting warnings and other oddities during compilation. I don't like warnings at any time, and with the kernel's make wrappers cleaning up the output they tend to stand out. Here's what I get: -- various type/attribute warnings -- reports of deprecated elements -- a report of "section mismatches", and instructions to use "make CONFIG_DEBUG_SECTION_MISMATCH=y" to find details. All that being said, the compilation completes, and I can boot it. I don't know the cause, but I have been unable to get vmware-server running on it, and I'm going back to the previous kernel for that reason. I hope you're not implying that a kernel should not be declared stable simply because a vmware product lags behind in kernel support? vmware 4 lacks support for kernel 2.6 alltogether, maybe we should only have 2.4 stable because of that.
[gentoo-user] Re: kernel 2.6.25-r6 oddities; is this kernel really ready for stable?
Kevin O'Gorman gmail.com> writes: > >> I run gentoo x86 stable, so that I usually avoid this sort of thing. > I am not suggesting that the vmware stuff should prevent releasing a > kernel. I am suggesting that pages of > cryptic and unusual output, not to mention warnings and deprecation > from the kernel's own make is cause for users to worry about the > kernel and wonder WTF. I run 2.6.25-gentoo-r6 on many amd64 machines, and have not noticed any significant issue or issues above the normal (as I perceive) expected results. Kernel hackers all assume we are their lab-rats, we exist for their tests (Kind of jaded but from a firmware engineer it is par for the coarse). This may be of little help to you, but, I fell better voicing my beliefs, because I spend countless hours parsing up and down device drivers and hardware issue, related to all kinds of firmware. So just ignore my noise, but, do understand that pleasing the masses is far from the goals of most kernel type hackers. The kernel is full of very bright people that are paid by corporations, whose goals are to make money. Jaded (YES) Accurate (YES) useful (?). James
[gentoo-user] Re: kernel 2.6.25-r6 oddities; is this kernel really ready for stable?
Dale wrote: You are not alone if having "issues" with this kernel. When I tried to run it recently, I noticed a serious slow down in KDE, especially when logging into KDE. My usual login time is about 7 to 8 seconds but with this kernel, try about 30 seconds. My mouse was "jerky" and even Seamonkey was very slow to switch tabs. Nothing changed but the kernel. Check if you're using a graphics driver that needs its modules recompiled after a kernel upgrade or else it either disables acceleration or falls back to the VESA or FB driver (slow). I didn't test to long because it was so annoying. I used make oldconfig and my old kernel config as a starting point. Did you happen to do the same? Maybe it should be done from scratch? Some conflict or something? Thoughts? I never needed to do 'make oldconfig'. After emerging a new kernel, I do "make menuconfig" (or xconfig (Qt) or gconfig (Gtk)) and everything is already configured like the currently running kernel. I think this requires the following options enabled though (General setup): <*> Kernel .config support [*] Enable access to .config through /proc/config.gz I think it's on by default.
[gentoo-user] Re: kernel 2.6.25-r6 oddities; is this kernel really ready for stable?
Kevin O'Gorman wrote: [...] BTW: I told a few other faculty about my difficulties: here's a typical reaction: It's an obvious question, but... are they freakin' insane? VMWare is one of the few really good pieces of general-market software that supports Linux. Why on earth would the linux community cut them off? Harumph! Mind you, I'm not certain who cut who off, but I'll be glad to report there's a serious effort to fix this. No one has cut off anyone. VMWare even contributes code to the kernel (and it is accepted). The problem is simply that VMWare chooses to support specific versions of the kernel. But when you ask them about support for the new kernels, the answer is "we are annoyed by the kernel development process; they change the internal API all the time, it's annoying." I think it's time for the kernel devs to apologize to VMWare and to promise that they won't be so annoying anymore in the future. We can't let VMWare getting annoyed, now can we.
Re: [gentoo-user] Re: kernel 2.6.25-r6 oddities; is this kernel really ready for stable?
On Wed, Jul 23, 2008 at 10:04 AM, Nikos Chantziaras <[EMAIL PROTECTED]> wrote: > Kevin O'Gorman wrote: >> >> I run gentoo x86 stable, so that I usually avoid this sort of thing. >> >> This kernel, however, looks balky to me, because it's reporting >> warnings and other oddities during compilation. I don't like warnings >> at any time, and with the kernel's make wrappers cleaning up the >> output they tend to stand out. >> >> Here's what I get: >> -- various type/attribute warnings >> -- reports of deprecated elements >> -- a report of "section mismatches", and instructions to use "make >> CONFIG_DEBUG_SECTION_MISMATCH=y" to find details. >> >> All that being said, the compilation completes, and I can boot it. I >> don't know the cause, but I have been unable to get vmware-server >> running on it, and I'm going back to the previous kernel for that >> reason. > > I hope you're not implying that a kernel should not be declared stable > simply because a vmware product lags behind in kernel support? vmware 4 > lacks support for kernel 2.6 alltogether, maybe we should only have 2.4 > stable because of that. I am not suggesting that the vmware stuff should prevent releasing a kernel. I am suggesting that pages of cryptic and unusual output, not to mention warnings and deprecation from the kernel's own make is cause for users to worry about the kernel and wonder WTF. ++ kevin -- Kevin O'Gorman, PhD
Re: [gentoo-user] Re: kernel 2.6.25-r6 oddities; is this kernel really ready for stable?
On Thu, 24 Jul 2008 03:55:14 +0300, Nikos Chantziaras wrote: > I never needed to do 'make oldconfig'. After emerging a new kernel, I > do "make menuconfig" (or xconfig (Qt) or gconfig (Gtk)) and everything > is already configured like the currently running kernel. But your new kernel is different from the current one, with different configuration options. oldconfig shows you the new options so you can choose which to set, whereas your method sets everything to the default. No, you don't strictly need to run make oldconfig, but it is advisable or you will be enabling options you didn't even know existed. -- Neil Bothwick The Japanese call us lazy, but at least we cook our fish! signature.asc Description: PGP signature
Re: [gentoo-user] Re: kernel 2.6.25-r6 oddities; is this kernel really ready for stable?
On Fri, Jul 25, 2008 at 7:18 AM, Nikos Chantziaras <[EMAIL PROTECTED]> wrote: > > > But when you ask them about support for the new kernels, the answer is "we > are annoyed by the kernel development process; they change the internal API > all the time, it's annoying." > > I think it's time for the kernel devs to apologize to VMWare and to promise > that they won't be so annoying anymore in the future. We can't let VMWare > getting annoyed, now can we. > > read "stable-api-nonsense.txt" in any kernel source directory. That's the way it works. If VMWare doesn't like it, that's tough. They DO have the option of putting their entire code base in the kernel tree like every other driver that is already there, but they have chosen not to. That's their choice and no-one forced them to do it. Now they have to live with that choice and stop whinging. Linux existed long before VMWare came along. It will likely still exist long after VMWare has gone away. -- Alan McKinnon [EMAIL PROTECTED]
Re: [gentoo-user] Re: kernel 2.6.25-r6 oddities; is this kernel really ready for stable?
2008/7/26 Alan McKinnon <[EMAIL PROTECTED]> > On Fri, Jul 25, 2008 at 7:18 AM, Nikos Chantziaras <[EMAIL PROTECTED]> > wrote: > >> >> >> But when you ask them about support for the new kernels, the answer is "we >> are annoyed by the kernel development process; they change the internal API >> all the time, it's annoying." >> >> I think it's time for the kernel devs to apologize to VMWare and to >> promise that they won't be so annoying anymore in the future. We can't let >> VMWare getting annoyed, now can we. >> >> > read "stable-api-nonsense.txt" in any kernel source directory. That's the > way it works. If VMWare doesn't like it, that's tough. They DO have the > option of putting their entire code base in the kernel tree like every other > driver that is already there, but they have chosen not to. That's their > choice and no-one forced them to do it. Now they have to live with that > choice and stop whinging. > > Linux existed long before VMWare came along. It will likely still exist > long after VMWare has gone away. > > > -- > Alan McKinnon > [EMAIL PROTECTED] > If you like to use VMWare, then you will have to like to use the kernel version that they support until they support a newer one... Then it's not about the stability of the gentoo-sources, it's about the VMWare people... There are new features in the new kernel that have been already tested in several distributions and that are important to keep gentoo up with the new hardware support I think... -- En el pasado creímos que se nos acababa el petróleo, pero en realidad lo que se nos acababan eran las IDEAS. http://www.lacomunidadpetrolera.com Gentoo/* (Linux #455615) www.gentoove.org 53:04:4e:b5:f9:7e:4a:fb:66:a2:19:12:da:d5:97:f8 8A3A C955 715D D88A 87CD E21F F827 ADD7 F589 B4A4 Sebastian Magrí (sebasmagri)