Re: [RFC PATCH 00/28] Linux Kernel Library
On Sat, Nov 7, 2015 at 12:48 PM, Richard W.M. Jones wrote: > > I just pushed a (very early) WIP branch that contains changes to > libguestfs to add an LKL backend: > > https://github.com/rwmjones/libguestfs/tree/lkl > > Read the README file in the libguestfs sources before starting, > followed by the instructions in the commit message: > > > https://github.com/rwmjones/libguestfs/commit/e38525f0b984d0a426f3348d95f2033673d4eaa4 > Hi Richard, Thanks for this, I was able to replicate it locally. I started working on adding a new library, liblkl-redirect.a, that redefines common libc symbols to use lkl. We then can link the daemon with this lib and lkl to avoid ugly ifdefs in the daemon code. Thanks, Tavi -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Sun, Nov 8, 2015 at 4:42 PM, yalin wang wrote: > > On Nov 4, 2015, at 07:06, Octavian Purdila > wrote: > > On Tue, Nov 3, 2015 at 11:40 PM, Richard Weinberger > wrote: > > Hi Richard, > > On Tue, Nov 3, 2015 at 9:20 PM, Octavian Purdila > wrote: > > LKL (Linux Kernel Library) is aiming to allow reusing the Linux kernel code > as extensively as possible with minimal effort and reduced maintenance > overhead. > > Examples of how LKL can be used are: creating userspace applications > (running on Linux and other operating systems) that can read or write Linux > filesystems or can use the Linux networking stack, creating kernel drivers > for other operating systems that can read Linux filesystems, bootloaders > support for reading/writing Linux filesystems, etc. > > With LKL, the kernel code is compiled into an object file that can be > directly linked by applications. The API offered by LKL is based on the > Linux system call interface. > > LKL is implemented as an architecture port in arch/lkl. It relies on host > operations defined by the application or a host library (tools/lkl/lib). > > The latest LKL version can be found at g...@github.com:lkl/linux.git > > > Or more copy friendly: https://github.com/lkl/linux.git > > FAQ > === > > Q: How is LKL different from UML? > A: UML provides a full OS environment (e.g. user/kernel separation, user > processes) and also has requirements (a filesystem, processes, etc.) that > makes it hard to use it for standalone applications. UML also relies > heavily on Linux hosts. On the other hand LKL is designed to be linked > directly with the application and hence does not have user/kernel > separation which makes it easier to use it in standalone applications. > > > So, this is a "liblinux" where applications are directly linked > against the kernel. > IOW system calls are plain function calls into the kernel? > > > More like "thread" calls. All system calls are executed in a dedicate > (kernel) thread to avoid race conditions with the "interrupt" path. > > why not call sys_XXX() function directly? > since kernel have implement lots of spin_locks to avoid race with normal > path > in IRQ handle , isn’t it ? > for example, you timer IRQ can be simulated by SIGALARM signal, > and the signal handler can check if IRQ is disabled , > if not , then continuing , otherwise , return directly .. > it is not safe ? > Hi Yalin, We need to have a proper Linux context in order to issue system calls (e.g. current needs to point to a proper Linux kernel thread_struct). We also want to let the Linux scheduler to select what kernel threads run at a given time. Lets say that currently a ksoftirqd runs and the application want to issue a system call. In this case we want to wait for ksoftirqd to complete then run the system call. The simplest way I found to do that is to have the system calls execute from Linux kernel threads hence the need to queue the system calls from the application thread to the Linux kernel system call thread. (BTW, we can have multiple system call threads if needed and the 2.6 implementation has that, but in order to simplify the review process I decided to throw away that for the moment). Thanks, Tavi -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Sun, Nov 8, 2015 at 4:42 PM, yalin wangwrote: > > On Nov 4, 2015, at 07:06, Octavian Purdila > wrote: > > On Tue, Nov 3, 2015 at 11:40 PM, Richard Weinberger > wrote: > > Hi Richard, > > On Tue, Nov 3, 2015 at 9:20 PM, Octavian Purdila > wrote: > > LKL (Linux Kernel Library) is aiming to allow reusing the Linux kernel code > as extensively as possible with minimal effort and reduced maintenance > overhead. > > Examples of how LKL can be used are: creating userspace applications > (running on Linux and other operating systems) that can read or write Linux > filesystems or can use the Linux networking stack, creating kernel drivers > for other operating systems that can read Linux filesystems, bootloaders > support for reading/writing Linux filesystems, etc. > > With LKL, the kernel code is compiled into an object file that can be > directly linked by applications. The API offered by LKL is based on the > Linux system call interface. > > LKL is implemented as an architecture port in arch/lkl. It relies on host > operations defined by the application or a host library (tools/lkl/lib). > > The latest LKL version can be found at g...@github.com:lkl/linux.git > > > Or more copy friendly: https://github.com/lkl/linux.git > > FAQ > === > > Q: How is LKL different from UML? > A: UML provides a full OS environment (e.g. user/kernel separation, user > processes) and also has requirements (a filesystem, processes, etc.) that > makes it hard to use it for standalone applications. UML also relies > heavily on Linux hosts. On the other hand LKL is designed to be linked > directly with the application and hence does not have user/kernel > separation which makes it easier to use it in standalone applications. > > > So, this is a "liblinux" where applications are directly linked > against the kernel. > IOW system calls are plain function calls into the kernel? > > > More like "thread" calls. All system calls are executed in a dedicate > (kernel) thread to avoid race conditions with the "interrupt" path. > > why not call sys_XXX() function directly? > since kernel have implement lots of spin_locks to avoid race with normal > path > in IRQ handle , isn’t it ? > for example, you timer IRQ can be simulated by SIGALARM signal, > and the signal handler can check if IRQ is disabled , > if not , then continuing , otherwise , return directly .. > it is not safe ? > Hi Yalin, We need to have a proper Linux context in order to issue system calls (e.g. current needs to point to a proper Linux kernel thread_struct). We also want to let the Linux scheduler to select what kernel threads run at a given time. Lets say that currently a ksoftirqd runs and the application want to issue a system call. In this case we want to wait for ksoftirqd to complete then run the system call. The simplest way I found to do that is to have the system calls execute from Linux kernel threads hence the need to queue the system calls from the application thread to the Linux kernel system call thread. (BTW, we can have multiple system call threads if needed and the 2.6 implementation has that, but in order to simplify the review process I decided to throw away that for the moment). Thanks, Tavi -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Sat, Nov 7, 2015 at 12:48 PM, Richard W.M. Joneswrote: > > I just pushed a (very early) WIP branch that contains changes to > libguestfs to add an LKL backend: > > https://github.com/rwmjones/libguestfs/tree/lkl > > Read the README file in the libguestfs sources before starting, > followed by the instructions in the commit message: > > > https://github.com/rwmjones/libguestfs/commit/e38525f0b984d0a426f3348d95f2033673d4eaa4 > Hi Richard, Thanks for this, I was able to replicate it locally. I started working on adding a new library, liblkl-redirect.a, that redefines common libc symbols to use lkl. We then can link the daemon with this lib and lkl to avoid ugly ifdefs in the daemon code. Thanks, Tavi -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
Hello Octavian, At Tue, 3 Nov 2015 22:20:31 +0200, Octavian Purdila wrote: > > > Q: How is LKL different from LibOS? > A: LibOS re-implements high-level kernel APIs for timers, softirqs, > scheduling, sysctl, SLAB/SLUB, etc. LKL behaves like any arch port, > implementing the arch level operations requested by the Linux kernel. LKL > also offers a host interface so that support for multiple hosts can be > easily implemented. I review most of code with the help of document and paper (2010). I think LKL and LibOS are essentially the same thing. I describe the current differences of both features, which I believe there are no fundamental ones (i.e., both can improve by putting some efforts). - LKL (beautiful) arch implementation (I like it) fully kbuild compliant rich fs support host support: POSIX, win, haiku, etc - LibOS existing application integration (semi-automated) system call table generation multiple process support (via system call proxy) various network backends (raw socket, DPDK, netmap, tap) symbol namespace separation host support: == rump hypercall (POSIX, xen, qemu/kvm/baremetal(under development)), ns-3 simulator # I can't find network support within the current patch but there is/will be a certain code that LKL can play with networking subsystem. existing application integration is really important when you want to configure network stack: since the configuration of file systems is just a mount(), but configurations of network stack need much userspace applications like iproute2 (ip, ss, tc) etc, which is not trivial to re-implement. -- Hajime -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
Hello Octavian, At Tue, 3 Nov 2015 22:20:31 +0200, Octavian Purdila wrote: > > > Q: How is LKL different from LibOS? > A: LibOS re-implements high-level kernel APIs for timers, softirqs, > scheduling, sysctl, SLAB/SLUB, etc. LKL behaves like any arch port, > implementing the arch level operations requested by the Linux kernel. LKL > also offers a host interface so that support for multiple hosts can be > easily implemented. I review most of code with the help of document and paper (2010). I think LKL and LibOS are essentially the same thing. I describe the current differences of both features, which I believe there are no fundamental ones (i.e., both can improve by putting some efforts). - LKL (beautiful) arch implementation (I like it) fully kbuild compliant rich fs support host support: POSIX, win, haiku, etc - LibOS existing application integration (semi-automated) system call table generation multiple process support (via system call proxy) various network backends (raw socket, DPDK, netmap, tap) symbol namespace separation host support: == rump hypercall (POSIX, xen, qemu/kvm/baremetal(under development)), ns-3 simulator # I can't find network support within the current patch but there is/will be a certain code that LKL can play with networking subsystem. existing application integration is really important when you want to configure network stack: since the configuration of file systems is just a mount(), but configurations of network stack need much userspace applications like iproute2 (ip, ss, tc) etc, which is not trivial to re-implement. -- Hajime -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Sat, Nov 7, 2015 at 2:35 AM, Richard Weinberger wrote: > Am 04.11.2015 um 15:15 schrieb Octavian Purdila: >> We could redefine the syscalls/libc symbols to call lkl_sys_ functions >> in launch-lkl, e.g.: >> >> int opendir(const char *path) >> { >>return lkl_opendir(new_path) >> } > > To get a better feeling how LKL behaves I've started with a tool > to mount any Linux filesystem by FUSE. > I.e. such that we can finally automount without root and bugs in filesystem > code won't hurt that much. > Hi Richard, I finished coding mine up yesterday :) I've just pushed it to github if you want to take a look at it (together with some small fixes). > lkl_sys_fstatat64() uses the type struct lkl_stat64. Where is it defined? > git grep is unable to locate it. > At least it seems to be incompatible with my local struct stat. > When doing the build in tools/lkl the LKL headers are installed in include/lkl. You should find it in there: $:~/src/linux/tools/lkl$ find include/ -type f | xargs grep lkl_stat64 include/lkl/asm-generic/stat.h:struct lkl_stat64 { > And why is there no lkl_sys_openat() syscall? > Didn't get to test it hence I didn't add it to the API yet, as many other system calls. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Sat, Nov 7, 2015 at 2:35 AM, Richard Weinberger wrote: > Am 04.11.2015 um 15:15 schrieb Octavian Purdila: >> We could redefine the syscalls/libc symbols to call lkl_sys_ functions >> in launch-lkl, e.g.: >> >> int opendir(const char *path) >> { >>return lkl_opendir(new_path) >> } > > To get a better feeling how LKL behaves I've started with a tool > to mount any Linux filesystem by FUSE. > I.e. such that we can finally automount without root and bugs in filesystem > code won't hurt that much. > > lkl_sys_fstatat64() uses the type struct lkl_stat64. Where is it defined? > git grep is unable to locate it. > At least it seems to be incompatible with my local struct stat. > > And why is there no lkl_sys_openat() syscall? > > Thanks, > //richard -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
I just pushed a (very early) WIP branch that contains changes to libguestfs to add an LKL backend: https://github.com/rwmjones/libguestfs/tree/lkl Read the README file in the libguestfs sources before starting, followed by the instructions in the commit message: https://github.com/rwmjones/libguestfs/commit/e38525f0b984d0a426f3348d95f2033673d4eaa4 Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com virt-df lists disk usage of guests without needing to install any software inside the virtual machine. Supports Linux and Windows. http://people.redhat.com/~rjones/virt-df/ -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
I just pushed a (very early) WIP branch that contains changes to libguestfs to add an LKL backend: https://github.com/rwmjones/libguestfs/tree/lkl Read the README file in the libguestfs sources before starting, followed by the instructions in the commit message: https://github.com/rwmjones/libguestfs/commit/e38525f0b984d0a426f3348d95f2033673d4eaa4 Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com virt-df lists disk usage of guests without needing to install any software inside the virtual machine. Supports Linux and Windows. http://people.redhat.com/~rjones/virt-df/ -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Sat, Nov 7, 2015 at 2:35 AM, Richard Weinbergerwrote: > Am 04.11.2015 um 15:15 schrieb Octavian Purdila: >> We could redefine the syscalls/libc symbols to call lkl_sys_ functions >> in launch-lkl, e.g.: >> >> int opendir(const char *path) >> { >>return lkl_opendir(new_path) >> } > > To get a better feeling how LKL behaves I've started with a tool > to mount any Linux filesystem by FUSE. > I.e. such that we can finally automount without root and bugs in filesystem > code won't hurt that much. > > lkl_sys_fstatat64() uses the type struct lkl_stat64. Where is it defined? > git grep is unable to locate it. > At least it seems to be incompatible with my local struct stat. > > And why is there no lkl_sys_openat() syscall? > > Thanks, > //richard -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Sat, Nov 7, 2015 at 2:35 AM, Richard Weinbergerwrote: > Am 04.11.2015 um 15:15 schrieb Octavian Purdila: >> We could redefine the syscalls/libc symbols to call lkl_sys_ functions >> in launch-lkl, e.g.: >> >> int opendir(const char *path) >> { >>return lkl_opendir(new_path) >> } > > To get a better feeling how LKL behaves I've started with a tool > to mount any Linux filesystem by FUSE. > I.e. such that we can finally automount without root and bugs in filesystem > code won't hurt that much. > Hi Richard, I finished coding mine up yesterday :) I've just pushed it to github if you want to take a look at it (together with some small fixes). > lkl_sys_fstatat64() uses the type struct lkl_stat64. Where is it defined? > git grep is unable to locate it. > At least it seems to be incompatible with my local struct stat. > When doing the build in tools/lkl the LKL headers are installed in include/lkl. You should find it in there: $:~/src/linux/tools/lkl$ find include/ -type f | xargs grep lkl_stat64 include/lkl/asm-generic/stat.h:struct lkl_stat64 { > And why is there no lkl_sys_openat() syscall? > Didn't get to test it hence I didn't add it to the API yet, as many other system calls. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Sat, Nov 07, 2015 at 01:35:36AM +0100, Richard Weinberger wrote: > Am 04.11.2015 um 15:15 schrieb Octavian Purdila: > > We could redefine the syscalls/libc symbols to call lkl_sys_ functions > > in launch-lkl, e.g.: > > > > int opendir(const char *path) > > { > >return lkl_opendir(new_path) > > } > > To get a better feeling how LKL behaves I've started with a tool > to mount any Linux filesystem by FUSE. > I.e. such that we can finally automount without root and bugs in filesystem > code won't hurt that much. guestmount already does this: http://libguestfs.org/guestmount.1.html By porting a small amount of code from the daemon/ directory, it could do it using lkl too. See: http://www.gossamer-threads.com/lists/linux/kernel/2296116#2296116 Rich. > lkl_sys_fstatat64() uses the type struct lkl_stat64. Where is it defined? > git grep is unable to locate it. > At least it seems to be incompatible with my local struct stat. > > And why is there no lkl_sys_openat() syscall? > > Thanks, > //richard -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com Fedora Windows cross-compiler. Compile Windows programs, test, and build Windows installers. Over 100 libraries supported. http://fedoraproject.org/wiki/MinGW -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
Am 04.11.2015 um 15:15 schrieb Octavian Purdila: > We could redefine the syscalls/libc symbols to call lkl_sys_ functions > in launch-lkl, e.g.: > > int opendir(const char *path) > { >return lkl_opendir(new_path) > } To get a better feeling how LKL behaves I've started with a tool to mount any Linux filesystem by FUSE. I.e. such that we can finally automount without root and bugs in filesystem code won't hurt that much. lkl_sys_fstatat64() uses the type struct lkl_stat64. Where is it defined? git grep is unable to locate it. At least it seems to be incompatible with my local struct stat. And why is there no lkl_sys_openat() syscall? Thanks, //richard -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
Am 04.11.2015 um 15:15 schrieb Octavian Purdila: > We could redefine the syscalls/libc symbols to call lkl_sys_ functions > in launch-lkl, e.g.: > > int opendir(const char *path) > { >return lkl_opendir(new_path) > } To get a better feeling how LKL behaves I've started with a tool to mount any Linux filesystem by FUSE. I.e. such that we can finally automount without root and bugs in filesystem code won't hurt that much. lkl_sys_fstatat64() uses the type struct lkl_stat64. Where is it defined? git grep is unable to locate it. At least it seems to be incompatible with my local struct stat. And why is there no lkl_sys_openat() syscall? Thanks, //richard -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Sat, Nov 07, 2015 at 01:35:36AM +0100, Richard Weinberger wrote: > Am 04.11.2015 um 15:15 schrieb Octavian Purdila: > > We could redefine the syscalls/libc symbols to call lkl_sys_ functions > > in launch-lkl, e.g.: > > > > int opendir(const char *path) > > { > >return lkl_opendir(new_path) > > } > > To get a better feeling how LKL behaves I've started with a tool > to mount any Linux filesystem by FUSE. > I.e. such that we can finally automount without root and bugs in filesystem > code won't hurt that much. guestmount already does this: http://libguestfs.org/guestmount.1.html By porting a small amount of code from the daemon/ directory, it could do it using lkl too. See: http://www.gossamer-threads.com/lists/linux/kernel/2296116#2296116 Rich. > lkl_sys_fstatat64() uses the type struct lkl_stat64. Where is it defined? > git grep is unable to locate it. > At least it seems to be incompatible with my local struct stat. > > And why is there no lkl_sys_openat() syscall? > > Thanks, > //richard -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com Fedora Windows cross-compiler. Compile Windows programs, test, and build Windows installers. Over 100 libraries supported. http://fedoraproject.org/wiki/MinGW -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Wed, Nov 4, 2015 at 3:50 PM, Richard W.M. Jones wrote: > On Wed, Nov 04, 2015 at 01:24:03AM +0200, Octavian Purdila wrote: >> Thanks for the pointers Richard, I am going to take a look at it. > > Now I've had a chance to look at some of the example LKL tools, here's > what this actually involves. It's not actually a great deal of work, > it could probably be done in a day or two, but see my question about > `lkl_sys_*' below. > > libguestfs (the library part) needs to talk over an RPC connection to > its daemon. See diagram here: > > http://libguestfs.org/guestfs-internals.1.html > > The code in src/launch-{direct,libvirt,uml,...}.c sets up that > connection and runs the daemon -- normally inside a qemu wrapper, but > it could be inside UML. For LKL I think it should just fork the > daemon directly. > > The daemon would then be linked to LKL. > > So really what's needed is a src/launch-lkl.c probably modelled after > one of these current backends: > > https://github.com/libguestfs/libguestfs/blob/master/src/launch-uml.c > https://github.com/libguestfs/libguestfs/blob/master/src/launch-unix.c > > and then recompile the daemon to link to LKL: > > https://github.com/libguestfs/libguestfs/tree/master/daemon > > and pass the list of disk images to the daemon, probably best to do > that on the guestfsd command line. > > My only problem here: you can't just link to daemon to LKL, do you > have to change all of the system calls from `foo' to `lkl_sys_foo'? > That's an awful lot of #ifdefs ... > We could redefine the syscalls/libc symbols to call lkl_sys_ functions in launch-lkl, e.g.: int opendir(const char *path) { return lkl_opendir(new_path) } -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Wed, Nov 04, 2015 at 01:24:03AM +0200, Octavian Purdila wrote: > Thanks for the pointers Richard, I am going to take a look at it. Now I've had a chance to look at some of the example LKL tools, here's what this actually involves. It's not actually a great deal of work, it could probably be done in a day or two, but see my question about `lkl_sys_*' below. libguestfs (the library part) needs to talk over an RPC connection to its daemon. See diagram here: http://libguestfs.org/guestfs-internals.1.html The code in src/launch-{direct,libvirt,uml,...}.c sets up that connection and runs the daemon -- normally inside a qemu wrapper, but it could be inside UML. For LKL I think it should just fork the daemon directly. The daemon would then be linked to LKL. So really what's needed is a src/launch-lkl.c probably modelled after one of these current backends: https://github.com/libguestfs/libguestfs/blob/master/src/launch-uml.c https://github.com/libguestfs/libguestfs/blob/master/src/launch-unix.c and then recompile the daemon to link to LKL: https://github.com/libguestfs/libguestfs/tree/master/daemon and pass the list of disk images to the daemon, probably best to do that on the guestfsd command line. My only problem here: you can't just link to daemon to LKL, do you have to change all of the system calls from `foo' to `lkl_sys_foo'? That's an awful lot of #ifdefs ... Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com virt-top is 'top' for virtual machines. Tiny program with many powerful monitoring features, net stats, disk stats, logging, etc. http://people.redhat.com/~rjones/virt-top -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On 2015-11-03 18:24, Octavian Purdila wrote: On Wed, Nov 4, 2015 at 12:45 AM, Richard W.M. Jones wrote: I'm dubious that a lib-based approach could support LVM, partioning, ntfs-3g, qcow2, vmdk and all the other libguestfs stuff that relies on userspace tools + qemu as well as just the kernel drivers. Nevertheless a fast subset of libguestfs supporting just kernel filesystem drivers could be useful. LKL uses the full Linux I/O stack and I think LVM and partitioning should work out of the box. Adding support for qcow2 and vmdk should be possible as well. ntfs-3g might be problematic. Partitioning will work fine based on what you say. MD using the old metadata and automatic assembly should also work fine (assuming there is some way to tell the library to simulate running initcalls). DM (and by extension LVM, which is a bunch of userspace stuff on top of regular DM) and new style MD both require userspace tools to configure and interact with, this could be handled in two different ways: 1. Update the LVM and MD tools so they can be built as libraries and work directly with LKL. 2. Provide some wrapper functions to emulate dmsetup, lvm, and mdadm as distinct library calls. Of these, the first option is likely to be the best for long-term support, but the second is probably going to be easier to code quickly. QCOW2 and VMDK are both VM disk formats, and while I would love to see a driver treat them (and VDI, and VHD) like regular disk images on Linux in general, that will take some effort to implement properly. NTFS-3G is a FUSE based filesystem driver, so that kind of functionality would probably need to be implemented in the application itself (although having some way to have the app just link to it instead would be absolutely wonderful). smime.p7s Description: S/MIME Cryptographic Signature
Re: [RFC PATCH 00/28] Linux Kernel Library
On Wed, Nov 4, 2015 at 3:50 PM, Richard W.M. Joneswrote: > On Wed, Nov 04, 2015 at 01:24:03AM +0200, Octavian Purdila wrote: >> Thanks for the pointers Richard, I am going to take a look at it. > > Now I've had a chance to look at some of the example LKL tools, here's > what this actually involves. It's not actually a great deal of work, > it could probably be done in a day or two, but see my question about > `lkl_sys_*' below. > > libguestfs (the library part) needs to talk over an RPC connection to > its daemon. See diagram here: > > http://libguestfs.org/guestfs-internals.1.html > > The code in src/launch-{direct,libvirt,uml,...}.c sets up that > connection and runs the daemon -- normally inside a qemu wrapper, but > it could be inside UML. For LKL I think it should just fork the > daemon directly. > > The daemon would then be linked to LKL. > > So really what's needed is a src/launch-lkl.c probably modelled after > one of these current backends: > > https://github.com/libguestfs/libguestfs/blob/master/src/launch-uml.c > https://github.com/libguestfs/libguestfs/blob/master/src/launch-unix.c > > and then recompile the daemon to link to LKL: > > https://github.com/libguestfs/libguestfs/tree/master/daemon > > and pass the list of disk images to the daemon, probably best to do > that on the guestfsd command line. > > My only problem here: you can't just link to daemon to LKL, do you > have to change all of the system calls from `foo' to `lkl_sys_foo'? > That's an awful lot of #ifdefs ... > We could redefine the syscalls/libc symbols to call lkl_sys_ functions in launch-lkl, e.g.: int opendir(const char *path) { return lkl_opendir(new_path) } -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On 2015-11-03 18:24, Octavian Purdila wrote: On Wed, Nov 4, 2015 at 12:45 AM, Richard W.M. Joneswrote: I'm dubious that a lib-based approach could support LVM, partioning, ntfs-3g, qcow2, vmdk and all the other libguestfs stuff that relies on userspace tools + qemu as well as just the kernel drivers. Nevertheless a fast subset of libguestfs supporting just kernel filesystem drivers could be useful. LKL uses the full Linux I/O stack and I think LVM and partitioning should work out of the box. Adding support for qcow2 and vmdk should be possible as well. ntfs-3g might be problematic. Partitioning will work fine based on what you say. MD using the old metadata and automatic assembly should also work fine (assuming there is some way to tell the library to simulate running initcalls). DM (and by extension LVM, which is a bunch of userspace stuff on top of regular DM) and new style MD both require userspace tools to configure and interact with, this could be handled in two different ways: 1. Update the LVM and MD tools so they can be built as libraries and work directly with LKL. 2. Provide some wrapper functions to emulate dmsetup, lvm, and mdadm as distinct library calls. Of these, the first option is likely to be the best for long-term support, but the second is probably going to be easier to code quickly. QCOW2 and VMDK are both VM disk formats, and while I would love to see a driver treat them (and VDI, and VHD) like regular disk images on Linux in general, that will take some effort to implement properly. NTFS-3G is a FUSE based filesystem driver, so that kind of functionality would probably need to be implemented in the application itself (although having some way to have the app just link to it instead would be absolutely wonderful). smime.p7s Description: S/MIME Cryptographic Signature
Re: [RFC PATCH 00/28] Linux Kernel Library
On Wed, Nov 04, 2015 at 01:24:03AM +0200, Octavian Purdila wrote: > Thanks for the pointers Richard, I am going to take a look at it. Now I've had a chance to look at some of the example LKL tools, here's what this actually involves. It's not actually a great deal of work, it could probably be done in a day or two, but see my question about `lkl_sys_*' below. libguestfs (the library part) needs to talk over an RPC connection to its daemon. See diagram here: http://libguestfs.org/guestfs-internals.1.html The code in src/launch-{direct,libvirt,uml,...}.c sets up that connection and runs the daemon -- normally inside a qemu wrapper, but it could be inside UML. For LKL I think it should just fork the daemon directly. The daemon would then be linked to LKL. So really what's needed is a src/launch-lkl.c probably modelled after one of these current backends: https://github.com/libguestfs/libguestfs/blob/master/src/launch-uml.c https://github.com/libguestfs/libguestfs/blob/master/src/launch-unix.c and then recompile the daemon to link to LKL: https://github.com/libguestfs/libguestfs/tree/master/daemon and pass the list of disk images to the daemon, probably best to do that on the guestfsd command line. My only problem here: you can't just link to daemon to LKL, do you have to change all of the system calls from `foo' to `lkl_sys_foo'? That's an awful lot of #ifdefs ... Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com virt-top is 'top' for virtual machines. Tiny program with many powerful monitoring features, net stats, disk stats, logging, etc. http://people.redhat.com/~rjones/virt-top -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Wed, Nov 4, 2015 at 12:45 AM, Richard W.M. Jones wrote: > On Tue, Nov 03, 2015 at 10:40:29PM +0100, Richard Weinberger wrote: >> On Tue, Nov 3, 2015 at 9:20 PM, Octavian Purdila >> wrote: >> > LKL (Linux Kernel Library) is aiming to allow reusing the Linux kernel code >> > as extensively as possible with minimal effort and reduced maintenance >> > overhead. >> > >> > Examples of how LKL can be used are: creating userspace applications >> > (running on Linux and other operating systems) that can read or write Linux >> > filesystems or can use the Linux networking stack, creating kernel drivers >> > for other operating systems that can read Linux filesystems, bootloaders >> > support for reading/writing Linux filesystems, etc. >> > >> > With LKL, the kernel code is compiled into an object file that can be >> > directly linked by applications. The API offered by LKL is based on the >> > Linux system call interface. >> > >> > LKL is implemented as an architecture port in arch/lkl. It relies on host >> > operations defined by the application or a host library (tools/lkl/lib). >> > >> > The latest LKL version can be found at g...@github.com:lkl/linux.git >> >> Or more copy friendly: https://github.com/lkl/linux.git >> >> > FAQ >> > === >> > >> > Q: How is LKL different from UML? >> > A: UML provides a full OS environment (e.g. user/kernel separation, user >> > processes) and also has requirements (a filesystem, processes, etc.) that >> > makes it hard to use it for standalone applications. UML also relies >> > heavily on Linux hosts. On the other hand LKL is designed to be linked >> > directly with the application and hence does not have user/kernel >> > separation which makes it easier to use it in standalone applications. >> >> So, this is a "liblinux" where applications are directly linked >> against the kernel. >> IOW system calls are plain function calls into the kernel? >> >> This eliminates UML's most problematic areas, system call handling via >> ptrace() >> and virtual memory management via SIGSEGV. :-) >> >> > Q: How is LKL different from LibOS? >> > A: LibOS re-implements high-level kernel APIs for timers, softirqs, >> > scheduling, sysctl, SLAB/SLUB, etc. LKL behaves like any arch port, >> > implementing the arch level operations requested by the Linux kernel. LKL >> > also offers a host interface so that support for multiple hosts can be >> > easily implemented. >> >> Yeah, these re-implementations are what I find most worrisome about LibOS. >> >> > >> > Building LKL the host library and LKL applications >> > == >> > >> > % cd tools/lkl >> > % make >> > >> > will build LKL as a object file, it will install it in tools/lkl/lib >> > together >> > with the headers files in tools/lkl/include then will build the host >> > library, >> > tests and a few of application examples: >> > >> > * tests/boot - a simple applications that uses LKL and exercises the basic >> > LKL APIs >> > >> > * fs2tar - a tool that converts a filesystem image to a tar archive >> > >> > * cptofs/cpfromfs - a tool that copies files to/from a filesystem image >> >> Seeing forward to have a libguestfs port. :-) > > Thanks - I was keeping an eye on libos (and on the NetBSD rump kernel > stuff before), ready to integrate them into libguestfs as soon as they > offered filesystem access. > > It's easy to write a libguestfs-compatible backend, which brings all > the virt-* tools from libguestfs to the new code. The UML one looks > like this: > > https://github.com/libguestfs/libguestfs/blob/master/src/launch-uml.c > Thanks for the pointers Richard, I am going to take a look at it. > > I'm dubious that a lib-based approach could support LVM, partioning, > ntfs-3g, qcow2, vmdk and all the other libguestfs stuff that relies on > userspace tools + qemu as well as just the kernel drivers. > Nevertheless a fast subset of libguestfs supporting just kernel > filesystem drivers could be useful. > LKL uses the full Linux I/O stack and I think LVM and partitioning should work out of the box. Adding support for qcow2 and vmdk should be possible as well. ntfs-3g might be problematic. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
At Tue, 3 Nov 2015 22:45:45 +, Richard W.M. Jones wrote: > > > * cptofs/cpfromfs - a tool that copies files to/from a filesystem image > > > > Seeing forward to have a libguestfs port. :-) > > Thanks - I was keeping an eye on libos (and on the NetBSD rump kernel > stuff before), ready to integrate them into libguestfs as soon as they > offered filesystem access. I've been working on fs support on libos recently during the integration with rump kernel _hypercall_, though it's still in the middle (but open(2) in a specific condition works fine at least). https://github.com/libos-nuse/net-next-nuse/tree/rump-hypcall I would expect to see more concrete patchset in near future. -- Hajime -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Tue, Nov 3, 2015 at 11:40 PM, Richard Weinberger wrote: Hi Richard, > On Tue, Nov 3, 2015 at 9:20 PM, Octavian Purdila > wrote: >> LKL (Linux Kernel Library) is aiming to allow reusing the Linux kernel code >> as extensively as possible with minimal effort and reduced maintenance >> overhead. >> >> Examples of how LKL can be used are: creating userspace applications >> (running on Linux and other operating systems) that can read or write Linux >> filesystems or can use the Linux networking stack, creating kernel drivers >> for other operating systems that can read Linux filesystems, bootloaders >> support for reading/writing Linux filesystems, etc. >> >> With LKL, the kernel code is compiled into an object file that can be >> directly linked by applications. The API offered by LKL is based on the >> Linux system call interface. >> >> LKL is implemented as an architecture port in arch/lkl. It relies on host >> operations defined by the application or a host library (tools/lkl/lib). >> >> The latest LKL version can be found at g...@github.com:lkl/linux.git > > Or more copy friendly: https://github.com/lkl/linux.git > >> FAQ >> === >> >> Q: How is LKL different from UML? >> A: UML provides a full OS environment (e.g. user/kernel separation, user >> processes) and also has requirements (a filesystem, processes, etc.) that >> makes it hard to use it for standalone applications. UML also relies >> heavily on Linux hosts. On the other hand LKL is designed to be linked >> directly with the application and hence does not have user/kernel >> separation which makes it easier to use it in standalone applications. > > So, this is a "liblinux" where applications are directly linked > against the kernel. > IOW system calls are plain function calls into the kernel? > More like "thread" calls. All system calls are executed in a dedicate (kernel) thread to avoid race conditions with the "interrupt" path. > This eliminates UML's most problematic areas, system call handling via > ptrace() > and virtual memory management via SIGSEGV. :-) > :) >> Q: How is LKL different from LibOS? >> A: LibOS re-implements high-level kernel APIs for timers, softirqs, >> scheduling, sysctl, SLAB/SLUB, etc. LKL behaves like any arch port, >> implementing the arch level operations requested by the Linux kernel. LKL >> also offers a host interface so that support for multiple hosts can be >> easily implemented. > > Yeah, these re-implementations are what I find most worrisome about LibOS. > >> >> Building LKL the host library and LKL applications >> == >> >> % cd tools/lkl >> % make >> >> will build LKL as a object file, it will install it in tools/lkl/lib together >> with the headers files in tools/lkl/include then will build the host library, >> tests and a few of application examples: >> >> * tests/boot - a simple applications that uses LKL and exercises the basic >> LKL APIs >> >> * fs2tar - a tool that converts a filesystem image to a tar archive >> >> * cptofs/cpfromfs - a tool that copies files to/from a filesystem image > > Seeing forward to have a libguestfs port. :-) > > Is LKL strictly single threaded? > At this point yes. SMP support is on my todo list though :) -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Tue, Nov 03, 2015 at 10:40:29PM +0100, Richard Weinberger wrote: > On Tue, Nov 3, 2015 at 9:20 PM, Octavian Purdila > wrote: > > LKL (Linux Kernel Library) is aiming to allow reusing the Linux kernel code > > as extensively as possible with minimal effort and reduced maintenance > > overhead. > > > > Examples of how LKL can be used are: creating userspace applications > > (running on Linux and other operating systems) that can read or write Linux > > filesystems or can use the Linux networking stack, creating kernel drivers > > for other operating systems that can read Linux filesystems, bootloaders > > support for reading/writing Linux filesystems, etc. > > > > With LKL, the kernel code is compiled into an object file that can be > > directly linked by applications. The API offered by LKL is based on the > > Linux system call interface. > > > > LKL is implemented as an architecture port in arch/lkl. It relies on host > > operations defined by the application or a host library (tools/lkl/lib). > > > > The latest LKL version can be found at g...@github.com:lkl/linux.git > > Or more copy friendly: https://github.com/lkl/linux.git > > > FAQ > > === > > > > Q: How is LKL different from UML? > > A: UML provides a full OS environment (e.g. user/kernel separation, user > > processes) and also has requirements (a filesystem, processes, etc.) that > > makes it hard to use it for standalone applications. UML also relies > > heavily on Linux hosts. On the other hand LKL is designed to be linked > > directly with the application and hence does not have user/kernel > > separation which makes it easier to use it in standalone applications. > > So, this is a "liblinux" where applications are directly linked > against the kernel. > IOW system calls are plain function calls into the kernel? > > This eliminates UML's most problematic areas, system call handling via > ptrace() > and virtual memory management via SIGSEGV. :-) > > > Q: How is LKL different from LibOS? > > A: LibOS re-implements high-level kernel APIs for timers, softirqs, > > scheduling, sysctl, SLAB/SLUB, etc. LKL behaves like any arch port, > > implementing the arch level operations requested by the Linux kernel. LKL > > also offers a host interface so that support for multiple hosts can be > > easily implemented. > > Yeah, these re-implementations are what I find most worrisome about LibOS. > > > > > Building LKL the host library and LKL applications > > == > > > > % cd tools/lkl > > % make > > > > will build LKL as a object file, it will install it in tools/lkl/lib > > together > > with the headers files in tools/lkl/include then will build the host > > library, > > tests and a few of application examples: > > > > * tests/boot - a simple applications that uses LKL and exercises the basic > > LKL APIs > > > > * fs2tar - a tool that converts a filesystem image to a tar archive > > > > * cptofs/cpfromfs - a tool that copies files to/from a filesystem image > > Seeing forward to have a libguestfs port. :-) Thanks - I was keeping an eye on libos (and on the NetBSD rump kernel stuff before), ready to integrate them into libguestfs as soon as they offered filesystem access. It's easy to write a libguestfs-compatible backend, which brings all the virt-* tools from libguestfs to the new code. The UML one looks like this: https://github.com/libguestfs/libguestfs/blob/master/src/launch-uml.c I'm dubious that a lib-based approach could support LVM, partioning, ntfs-3g, qcow2, vmdk and all the other libguestfs stuff that relies on userspace tools + qemu as well as just the kernel drivers. Nevertheless a fast subset of libguestfs supporting just kernel filesystem drivers could be useful. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com virt-top is 'top' for virtual machines. Tiny program with many powerful monitoring features, net stats, disk stats, logging, etc. http://people.redhat.com/~rjones/virt-top -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Tue, Nov 3, 2015 at 9:20 PM, Octavian Purdila wrote: > LKL (Linux Kernel Library) is aiming to allow reusing the Linux kernel code > as extensively as possible with minimal effort and reduced maintenance > overhead. > > Examples of how LKL can be used are: creating userspace applications > (running on Linux and other operating systems) that can read or write Linux > filesystems or can use the Linux networking stack, creating kernel drivers > for other operating systems that can read Linux filesystems, bootloaders > support for reading/writing Linux filesystems, etc. > > With LKL, the kernel code is compiled into an object file that can be > directly linked by applications. The API offered by LKL is based on the > Linux system call interface. > > LKL is implemented as an architecture port in arch/lkl. It relies on host > operations defined by the application or a host library (tools/lkl/lib). > > The latest LKL version can be found at g...@github.com:lkl/linux.git Or more copy friendly: https://github.com/lkl/linux.git > FAQ > === > > Q: How is LKL different from UML? > A: UML provides a full OS environment (e.g. user/kernel separation, user > processes) and also has requirements (a filesystem, processes, etc.) that > makes it hard to use it for standalone applications. UML also relies > heavily on Linux hosts. On the other hand LKL is designed to be linked > directly with the application and hence does not have user/kernel > separation which makes it easier to use it in standalone applications. So, this is a "liblinux" where applications are directly linked against the kernel. IOW system calls are plain function calls into the kernel? This eliminates UML's most problematic areas, system call handling via ptrace() and virtual memory management via SIGSEGV. :-) > Q: How is LKL different from LibOS? > A: LibOS re-implements high-level kernel APIs for timers, softirqs, > scheduling, sysctl, SLAB/SLUB, etc. LKL behaves like any arch port, > implementing the arch level operations requested by the Linux kernel. LKL > also offers a host interface so that support for multiple hosts can be > easily implemented. Yeah, these re-implementations are what I find most worrisome about LibOS. > > Building LKL the host library and LKL applications > == > > % cd tools/lkl > % make > > will build LKL as a object file, it will install it in tools/lkl/lib together > with the headers files in tools/lkl/include then will build the host library, > tests and a few of application examples: > > * tests/boot - a simple applications that uses LKL and exercises the basic > LKL APIs > > * fs2tar - a tool that converts a filesystem image to a tar archive > > * cptofs/cpfromfs - a tool that copies files to/from a filesystem image Seeing forward to have a libguestfs port. :-) Is LKL strictly single threaded? -- Thanks, //richard -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
[RFC PATCH 00/28] Linux Kernel Library
LKL (Linux Kernel Library) is aiming to allow reusing the Linux kernel code as extensively as possible with minimal effort and reduced maintenance overhead. Examples of how LKL can be used are: creating userspace applications (running on Linux and other operating systems) that can read or write Linux filesystems or can use the Linux networking stack, creating kernel drivers for other operating systems that can read Linux filesystems, bootloaders support for reading/writing Linux filesystems, etc. With LKL, the kernel code is compiled into an object file that can be directly linked by applications. The API offered by LKL is based on the Linux system call interface. LKL is implemented as an architecture port in arch/lkl. It relies on host operations defined by the application or a host library (tools/lkl/lib). The latest LKL version can be found at g...@github.com:lkl/linux.git FAQ === Q: How is LKL different from UML? A: UML provides a full OS environment (e.g. user/kernel separation, user processes) and also has requirements (a filesystem, processes, etc.) that makes it hard to use it for standalone applications. UML also relies heavily on Linux hosts. On the other hand LKL is designed to be linked directly with the application and hence does not have user/kernel separation which makes it easier to use it in standalone applications. Q: How is LKL different from LibOS? A: LibOS re-implements high-level kernel APIs for timers, softirqs, scheduling, sysctl, SLAB/SLUB, etc. LKL behaves like any arch port, implementing the arch level operations requested by the Linux kernel. LKL also offers a host interface so that support for multiple hosts can be easily implemented. Building LKL the host library and LKL applications == % cd tools/lkl % make will build LKL as a object file, it will install it in tools/lkl/lib together with the headers files in tools/lkl/include then will build the host library, tests and a few of application examples: * tests/boot - a simple applications that uses LKL and exercises the basic LKL APIs * fs2tar - a tool that converts a filesystem image to a tar archive * cptofs/cpfromfs - a tool that copies files to/from a filesystem image Supported hosts === Currently LKL supports POSIX and Windows userspace applications. New hosts can be added relatively easy if the host supports gcc and GNU ld. Previous versions of LKL supported Windows kernel and Haiku kernel hosts. Octavian Purdila (28): asm-generic: atomic64: allow using generic atomic64 on 64bit platforms kbuild: allow architectures to automatically define kconfig symbols lkl: architecture skeleton for Linux kernel library lkl: host interface lkl: memory handling lkl: kernel threads support lkl: interrupt support lkl: system call interface and application API lkl: timers, time and delay support lkl: memory mapped I/O support lkl: basic kernel console support init: allow architecture code to overide run_init_process lkl: initialization and cleanup lkl: plug in the build system lkl tools: skeleton for host side library, tests and tools lkl tools: host lib: add lkl_strerror and lkl_printf lkl tools: host lib: memory mapped I/O helpers lkl tools: host lib: virtio devices lkl tools: host lib: virtio block device lkl tools: host lib: filesystem helpers lkl tools: host lib: posix host operations lkl tools: "boot" test lkl tools: tool that converts a filesystem image to tar lkl tools: tool that reads/writes to/from a filesystem image signal: use CONFIG_X86_32 instead of __i386__ asm-generic: vmlinux.lds.h: allow customized rodata section name lkl: add support for Windows hosts lkl tools: add support for Windows host MAINTAINERS | 6 + Makefile| 1 + arch/lkl/.gitignore | 1 + arch/lkl/Kconfig| 83 ++ arch/lkl/Makefile | 39 +++ arch/lkl/auto.conf | 1 + arch/lkl/defconfig | 35 +++ arch/lkl/include/asm/Kbuild | 77 + arch/lkl/include/asm/bitsperlong.h | 11 + arch/lkl/include/asm/byteorder.h| 10 + arch/lkl/include/asm/dma-mapping.h | 6 + arch/lkl/include/asm/elf.h | 13 + arch/lkl/include/asm/host_ops.h | 9 + arch/lkl/include/asm/io.h | 104 +++ arch/lkl/include/asm/irq.h | 10 + arch/lkl/include/asm/mutex.h| 7 + arch/lkl/include/asm/page.h | 13 + arch/lkl/include/asm/pgtable.h | 60 arch/lkl/include/asm/processor.h| 53 arch/lkl/include/asm/ptrace.h | 23 ++ arch/lkl/include/asm/setup.h| 12 + arch/lkl/include/asm/thread_info.h | 82 + arch/lkl/include/asm/unistd.h | 93 ++ arch/lkl/include/asm/vmlinux.lds.h | 10 +
Re: [RFC PATCH 00/28] Linux Kernel Library
On Tue, Nov 3, 2015 at 9:20 PM, Octavian Purdilawrote: > LKL (Linux Kernel Library) is aiming to allow reusing the Linux kernel code > as extensively as possible with minimal effort and reduced maintenance > overhead. > > Examples of how LKL can be used are: creating userspace applications > (running on Linux and other operating systems) that can read or write Linux > filesystems or can use the Linux networking stack, creating kernel drivers > for other operating systems that can read Linux filesystems, bootloaders > support for reading/writing Linux filesystems, etc. > > With LKL, the kernel code is compiled into an object file that can be > directly linked by applications. The API offered by LKL is based on the > Linux system call interface. > > LKL is implemented as an architecture port in arch/lkl. It relies on host > operations defined by the application or a host library (tools/lkl/lib). > > The latest LKL version can be found at g...@github.com:lkl/linux.git Or more copy friendly: https://github.com/lkl/linux.git > FAQ > === > > Q: How is LKL different from UML? > A: UML provides a full OS environment (e.g. user/kernel separation, user > processes) and also has requirements (a filesystem, processes, etc.) that > makes it hard to use it for standalone applications. UML also relies > heavily on Linux hosts. On the other hand LKL is designed to be linked > directly with the application and hence does not have user/kernel > separation which makes it easier to use it in standalone applications. So, this is a "liblinux" where applications are directly linked against the kernel. IOW system calls are plain function calls into the kernel? This eliminates UML's most problematic areas, system call handling via ptrace() and virtual memory management via SIGSEGV. :-) > Q: How is LKL different from LibOS? > A: LibOS re-implements high-level kernel APIs for timers, softirqs, > scheduling, sysctl, SLAB/SLUB, etc. LKL behaves like any arch port, > implementing the arch level operations requested by the Linux kernel. LKL > also offers a host interface so that support for multiple hosts can be > easily implemented. Yeah, these re-implementations are what I find most worrisome about LibOS. > > Building LKL the host library and LKL applications > == > > % cd tools/lkl > % make > > will build LKL as a object file, it will install it in tools/lkl/lib together > with the headers files in tools/lkl/include then will build the host library, > tests and a few of application examples: > > * tests/boot - a simple applications that uses LKL and exercises the basic > LKL APIs > > * fs2tar - a tool that converts a filesystem image to a tar archive > > * cptofs/cpfromfs - a tool that copies files to/from a filesystem image Seeing forward to have a libguestfs port. :-) Is LKL strictly single threaded? -- Thanks, //richard -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Tue, Nov 3, 2015 at 11:40 PM, Richard Weinbergerwrote: Hi Richard, > On Tue, Nov 3, 2015 at 9:20 PM, Octavian Purdila > wrote: >> LKL (Linux Kernel Library) is aiming to allow reusing the Linux kernel code >> as extensively as possible with minimal effort and reduced maintenance >> overhead. >> >> Examples of how LKL can be used are: creating userspace applications >> (running on Linux and other operating systems) that can read or write Linux >> filesystems or can use the Linux networking stack, creating kernel drivers >> for other operating systems that can read Linux filesystems, bootloaders >> support for reading/writing Linux filesystems, etc. >> >> With LKL, the kernel code is compiled into an object file that can be >> directly linked by applications. The API offered by LKL is based on the >> Linux system call interface. >> >> LKL is implemented as an architecture port in arch/lkl. It relies on host >> operations defined by the application or a host library (tools/lkl/lib). >> >> The latest LKL version can be found at g...@github.com:lkl/linux.git > > Or more copy friendly: https://github.com/lkl/linux.git > >> FAQ >> === >> >> Q: How is LKL different from UML? >> A: UML provides a full OS environment (e.g. user/kernel separation, user >> processes) and also has requirements (a filesystem, processes, etc.) that >> makes it hard to use it for standalone applications. UML also relies >> heavily on Linux hosts. On the other hand LKL is designed to be linked >> directly with the application and hence does not have user/kernel >> separation which makes it easier to use it in standalone applications. > > So, this is a "liblinux" where applications are directly linked > against the kernel. > IOW system calls are plain function calls into the kernel? > More like "thread" calls. All system calls are executed in a dedicate (kernel) thread to avoid race conditions with the "interrupt" path. > This eliminates UML's most problematic areas, system call handling via > ptrace() > and virtual memory management via SIGSEGV. :-) > :) >> Q: How is LKL different from LibOS? >> A: LibOS re-implements high-level kernel APIs for timers, softirqs, >> scheduling, sysctl, SLAB/SLUB, etc. LKL behaves like any arch port, >> implementing the arch level operations requested by the Linux kernel. LKL >> also offers a host interface so that support for multiple hosts can be >> easily implemented. > > Yeah, these re-implementations are what I find most worrisome about LibOS. > >> >> Building LKL the host library and LKL applications >> == >> >> % cd tools/lkl >> % make >> >> will build LKL as a object file, it will install it in tools/lkl/lib together >> with the headers files in tools/lkl/include then will build the host library, >> tests and a few of application examples: >> >> * tests/boot - a simple applications that uses LKL and exercises the basic >> LKL APIs >> >> * fs2tar - a tool that converts a filesystem image to a tar archive >> >> * cptofs/cpfromfs - a tool that copies files to/from a filesystem image > > Seeing forward to have a libguestfs port. :-) > > Is LKL strictly single threaded? > At this point yes. SMP support is on my todo list though :) -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
On Tue, Nov 03, 2015 at 10:40:29PM +0100, Richard Weinberger wrote: > On Tue, Nov 3, 2015 at 9:20 PM, Octavian Purdila >wrote: > > LKL (Linux Kernel Library) is aiming to allow reusing the Linux kernel code > > as extensively as possible with minimal effort and reduced maintenance > > overhead. > > > > Examples of how LKL can be used are: creating userspace applications > > (running on Linux and other operating systems) that can read or write Linux > > filesystems or can use the Linux networking stack, creating kernel drivers > > for other operating systems that can read Linux filesystems, bootloaders > > support for reading/writing Linux filesystems, etc. > > > > With LKL, the kernel code is compiled into an object file that can be > > directly linked by applications. The API offered by LKL is based on the > > Linux system call interface. > > > > LKL is implemented as an architecture port in arch/lkl. It relies on host > > operations defined by the application or a host library (tools/lkl/lib). > > > > The latest LKL version can be found at g...@github.com:lkl/linux.git > > Or more copy friendly: https://github.com/lkl/linux.git > > > FAQ > > === > > > > Q: How is LKL different from UML? > > A: UML provides a full OS environment (e.g. user/kernel separation, user > > processes) and also has requirements (a filesystem, processes, etc.) that > > makes it hard to use it for standalone applications. UML also relies > > heavily on Linux hosts. On the other hand LKL is designed to be linked > > directly with the application and hence does not have user/kernel > > separation which makes it easier to use it in standalone applications. > > So, this is a "liblinux" where applications are directly linked > against the kernel. > IOW system calls are plain function calls into the kernel? > > This eliminates UML's most problematic areas, system call handling via > ptrace() > and virtual memory management via SIGSEGV. :-) > > > Q: How is LKL different from LibOS? > > A: LibOS re-implements high-level kernel APIs for timers, softirqs, > > scheduling, sysctl, SLAB/SLUB, etc. LKL behaves like any arch port, > > implementing the arch level operations requested by the Linux kernel. LKL > > also offers a host interface so that support for multiple hosts can be > > easily implemented. > > Yeah, these re-implementations are what I find most worrisome about LibOS. > > > > > Building LKL the host library and LKL applications > > == > > > > % cd tools/lkl > > % make > > > > will build LKL as a object file, it will install it in tools/lkl/lib > > together > > with the headers files in tools/lkl/include then will build the host > > library, > > tests and a few of application examples: > > > > * tests/boot - a simple applications that uses LKL and exercises the basic > > LKL APIs > > > > * fs2tar - a tool that converts a filesystem image to a tar archive > > > > * cptofs/cpfromfs - a tool that copies files to/from a filesystem image > > Seeing forward to have a libguestfs port. :-) Thanks - I was keeping an eye on libos (and on the NetBSD rump kernel stuff before), ready to integrate them into libguestfs as soon as they offered filesystem access. It's easy to write a libguestfs-compatible backend, which brings all the virt-* tools from libguestfs to the new code. The UML one looks like this: https://github.com/libguestfs/libguestfs/blob/master/src/launch-uml.c I'm dubious that a lib-based approach could support LVM, partioning, ntfs-3g, qcow2, vmdk and all the other libguestfs stuff that relies on userspace tools + qemu as well as just the kernel drivers. Nevertheless a fast subset of libguestfs supporting just kernel filesystem drivers could be useful. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com virt-top is 'top' for virtual machines. Tiny program with many powerful monitoring features, net stats, disk stats, logging, etc. http://people.redhat.com/~rjones/virt-top -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
[RFC PATCH 00/28] Linux Kernel Library
LKL (Linux Kernel Library) is aiming to allow reusing the Linux kernel code as extensively as possible with minimal effort and reduced maintenance overhead. Examples of how LKL can be used are: creating userspace applications (running on Linux and other operating systems) that can read or write Linux filesystems or can use the Linux networking stack, creating kernel drivers for other operating systems that can read Linux filesystems, bootloaders support for reading/writing Linux filesystems, etc. With LKL, the kernel code is compiled into an object file that can be directly linked by applications. The API offered by LKL is based on the Linux system call interface. LKL is implemented as an architecture port in arch/lkl. It relies on host operations defined by the application or a host library (tools/lkl/lib). The latest LKL version can be found at g...@github.com:lkl/linux.git FAQ === Q: How is LKL different from UML? A: UML provides a full OS environment (e.g. user/kernel separation, user processes) and also has requirements (a filesystem, processes, etc.) that makes it hard to use it for standalone applications. UML also relies heavily on Linux hosts. On the other hand LKL is designed to be linked directly with the application and hence does not have user/kernel separation which makes it easier to use it in standalone applications. Q: How is LKL different from LibOS? A: LibOS re-implements high-level kernel APIs for timers, softirqs, scheduling, sysctl, SLAB/SLUB, etc. LKL behaves like any arch port, implementing the arch level operations requested by the Linux kernel. LKL also offers a host interface so that support for multiple hosts can be easily implemented. Building LKL the host library and LKL applications == % cd tools/lkl % make will build LKL as a object file, it will install it in tools/lkl/lib together with the headers files in tools/lkl/include then will build the host library, tests and a few of application examples: * tests/boot - a simple applications that uses LKL and exercises the basic LKL APIs * fs2tar - a tool that converts a filesystem image to a tar archive * cptofs/cpfromfs - a tool that copies files to/from a filesystem image Supported hosts === Currently LKL supports POSIX and Windows userspace applications. New hosts can be added relatively easy if the host supports gcc and GNU ld. Previous versions of LKL supported Windows kernel and Haiku kernel hosts. Octavian Purdila (28): asm-generic: atomic64: allow using generic atomic64 on 64bit platforms kbuild: allow architectures to automatically define kconfig symbols lkl: architecture skeleton for Linux kernel library lkl: host interface lkl: memory handling lkl: kernel threads support lkl: interrupt support lkl: system call interface and application API lkl: timers, time and delay support lkl: memory mapped I/O support lkl: basic kernel console support init: allow architecture code to overide run_init_process lkl: initialization and cleanup lkl: plug in the build system lkl tools: skeleton for host side library, tests and tools lkl tools: host lib: add lkl_strerror and lkl_printf lkl tools: host lib: memory mapped I/O helpers lkl tools: host lib: virtio devices lkl tools: host lib: virtio block device lkl tools: host lib: filesystem helpers lkl tools: host lib: posix host operations lkl tools: "boot" test lkl tools: tool that converts a filesystem image to tar lkl tools: tool that reads/writes to/from a filesystem image signal: use CONFIG_X86_32 instead of __i386__ asm-generic: vmlinux.lds.h: allow customized rodata section name lkl: add support for Windows hosts lkl tools: add support for Windows host MAINTAINERS | 6 + Makefile| 1 + arch/lkl/.gitignore | 1 + arch/lkl/Kconfig| 83 ++ arch/lkl/Makefile | 39 +++ arch/lkl/auto.conf | 1 + arch/lkl/defconfig | 35 +++ arch/lkl/include/asm/Kbuild | 77 + arch/lkl/include/asm/bitsperlong.h | 11 + arch/lkl/include/asm/byteorder.h| 10 + arch/lkl/include/asm/dma-mapping.h | 6 + arch/lkl/include/asm/elf.h | 13 + arch/lkl/include/asm/host_ops.h | 9 + arch/lkl/include/asm/io.h | 104 +++ arch/lkl/include/asm/irq.h | 10 + arch/lkl/include/asm/mutex.h| 7 + arch/lkl/include/asm/page.h | 13 + arch/lkl/include/asm/pgtable.h | 60 arch/lkl/include/asm/processor.h| 53 arch/lkl/include/asm/ptrace.h | 23 ++ arch/lkl/include/asm/setup.h| 12 + arch/lkl/include/asm/thread_info.h | 82 + arch/lkl/include/asm/unistd.h | 93 ++ arch/lkl/include/asm/vmlinux.lds.h | 10 +
Re: [RFC PATCH 00/28] Linux Kernel Library
On Wed, Nov 4, 2015 at 12:45 AM, Richard W.M. Joneswrote: > On Tue, Nov 03, 2015 at 10:40:29PM +0100, Richard Weinberger wrote: >> On Tue, Nov 3, 2015 at 9:20 PM, Octavian Purdila >> wrote: >> > LKL (Linux Kernel Library) is aiming to allow reusing the Linux kernel code >> > as extensively as possible with minimal effort and reduced maintenance >> > overhead. >> > >> > Examples of how LKL can be used are: creating userspace applications >> > (running on Linux and other operating systems) that can read or write Linux >> > filesystems or can use the Linux networking stack, creating kernel drivers >> > for other operating systems that can read Linux filesystems, bootloaders >> > support for reading/writing Linux filesystems, etc. >> > >> > With LKL, the kernel code is compiled into an object file that can be >> > directly linked by applications. The API offered by LKL is based on the >> > Linux system call interface. >> > >> > LKL is implemented as an architecture port in arch/lkl. It relies on host >> > operations defined by the application or a host library (tools/lkl/lib). >> > >> > The latest LKL version can be found at g...@github.com:lkl/linux.git >> >> Or more copy friendly: https://github.com/lkl/linux.git >> >> > FAQ >> > === >> > >> > Q: How is LKL different from UML? >> > A: UML provides a full OS environment (e.g. user/kernel separation, user >> > processes) and also has requirements (a filesystem, processes, etc.) that >> > makes it hard to use it for standalone applications. UML also relies >> > heavily on Linux hosts. On the other hand LKL is designed to be linked >> > directly with the application and hence does not have user/kernel >> > separation which makes it easier to use it in standalone applications. >> >> So, this is a "liblinux" where applications are directly linked >> against the kernel. >> IOW system calls are plain function calls into the kernel? >> >> This eliminates UML's most problematic areas, system call handling via >> ptrace() >> and virtual memory management via SIGSEGV. :-) >> >> > Q: How is LKL different from LibOS? >> > A: LibOS re-implements high-level kernel APIs for timers, softirqs, >> > scheduling, sysctl, SLAB/SLUB, etc. LKL behaves like any arch port, >> > implementing the arch level operations requested by the Linux kernel. LKL >> > also offers a host interface so that support for multiple hosts can be >> > easily implemented. >> >> Yeah, these re-implementations are what I find most worrisome about LibOS. >> >> > >> > Building LKL the host library and LKL applications >> > == >> > >> > % cd tools/lkl >> > % make >> > >> > will build LKL as a object file, it will install it in tools/lkl/lib >> > together >> > with the headers files in tools/lkl/include then will build the host >> > library, >> > tests and a few of application examples: >> > >> > * tests/boot - a simple applications that uses LKL and exercises the basic >> > LKL APIs >> > >> > * fs2tar - a tool that converts a filesystem image to a tar archive >> > >> > * cptofs/cpfromfs - a tool that copies files to/from a filesystem image >> >> Seeing forward to have a libguestfs port. :-) > > Thanks - I was keeping an eye on libos (and on the NetBSD rump kernel > stuff before), ready to integrate them into libguestfs as soon as they > offered filesystem access. > > It's easy to write a libguestfs-compatible backend, which brings all > the virt-* tools from libguestfs to the new code. The UML one looks > like this: > > https://github.com/libguestfs/libguestfs/blob/master/src/launch-uml.c > Thanks for the pointers Richard, I am going to take a look at it. > > I'm dubious that a lib-based approach could support LVM, partioning, > ntfs-3g, qcow2, vmdk and all the other libguestfs stuff that relies on > userspace tools + qemu as well as just the kernel drivers. > Nevertheless a fast subset of libguestfs supporting just kernel > filesystem drivers could be useful. > LKL uses the full Linux I/O stack and I think LVM and partitioning should work out of the box. Adding support for qcow2 and vmdk should be possible as well. ntfs-3g might be problematic. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC PATCH 00/28] Linux Kernel Library
At Tue, 3 Nov 2015 22:45:45 +, Richard W.M. Jones wrote: > > > * cptofs/cpfromfs - a tool that copies files to/from a filesystem image > > > > Seeing forward to have a libguestfs port. :-) > > Thanks - I was keeping an eye on libos (and on the NetBSD rump kernel > stuff before), ready to integrate them into libguestfs as soon as they > offered filesystem access. I've been working on fs support on libos recently during the integration with rump kernel _hypercall_, though it's still in the middle (but open(2) in a specific condition works fine at least). https://github.com/libos-nuse/net-next-nuse/tree/rump-hypcall I would expect to see more concrete patchset in near future. -- Hajime -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/