Virtual device driver programming
Hi all, I'm interested in programming a virtual network device driver -Linux concept- on a FreeBSD box. The idea behind of this sort of interface is the following: From the kernel's point of view, a network interface is a software object that can process outgoing packets, and the actual transmission mechanism remains hidden inside the interface driver. Even though most interfaces are associated to physical devices (or, for the loopback interface, to a software-only data loop), it is possible to design network interface drivers that rely on other interfaces to perform actual packet transmission. The idea of a ``virtual'' interface can be useful to implement special-purpose processing on data packets while avoiding to hack with the network subsystem of the kernel. I know that is not too complicated to program this sort of functionality in linux but I would like to do it over freebsd, has anybody some idea on how could I start doing it? I've been searching in the freebsd handbook but I haven't found anything really relevant. Regards, Johnny PS: You can find more information about virtual network device driver in the following link: http://www.linux.it/~rubini/docs/vinter/vinter.html ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Virtual device driver programming
Hello! On Fri, 23 Jun 2006, Johnny Choque wrote: I'm interested in programming a virtual network device driver -Linux concept- on a FreeBSD box. The idea behind of this sort of interface is the following: From the kernel's point of view, a network interface is a software object that can process outgoing packets, and the actual transmission mechanism remains hidden inside the interface driver. Even though most interfaces are associated to physical devices (or, for the loopback interface, to a software-only data loop), it is possible to design network interface drivers that rely on other interfaces to perform actual packet transmission. The idea of a ``virtual'' interface can be useful to implement special-purpose processing on data packets while avoiding to hack with the network subsystem of the kernel. Isn't this exactly the same thing that FreeBSD's Netgraph subsystem does? man 4 netgraph Sincerely, Dmitry -- Atlantis ISP, System Administrator e-mail: [EMAIL PROTECTED] nic-hdl: LYNX-RIPE ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Virtual device driver programming
On Fri, 23 Jun 2006, Johnny Choque wrote: I'm interested in programming a virtual network device driver -Linux concept- on a FreeBSD box. The idea behind of this sort of interface is the following: From the kernel's point of view, a network interface is a software object that can process outgoing packets, and the actual transmission mechanism remains hidden inside the interface driver. Even though most interfaces are associated to physical devices (or, for the loopback interface, to a software-only data loop), it is possible to design network interface drivers that rely on other interfaces to perform actual packet transmission. The idea of a ``virtual'' interface can be useful to implement special-purpose processing on data packets while avoiding to hack with the network subsystem of the kernel. I know that is not too complicated to program this sort of functionality in linux but I would like to do it over freebsd, has anybody some idea on how could I start doing it? I've been searching in the freebsd handbook but I haven't found anything really relevant. tap(4) and tun(4) describe pseudo-devices you can use to instantiate ethernet and tunnel interfaces from user space. Programs attach to pseudo-devices, and using read/write operations on the pseudo-device, can receive and generate packets on the network interface. In kernel, the ifnet(9) API is used to implement network interfaces -- nothing in the API requires that the under-side of a network interface be hardware. In fact, a great many network types without underlying hardware have been implemented, including the loopback interface, encapsulation interfaces, and the tap/tun interface drivers. Robert N M Watson Computer Laboratory University of Cambridge ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to [EMAIL PROTECTED]
Simple question about mmap() system call
Hello! I'm writing an utility that should examine some bytes of a large file and modify them - that't all. I've decided to mmap() the file: void *diskp; if ((fd=open(argv[1], O_RDWR)) == -1) err(EX_NOINPUT, Can't open %s for readind and writing, argv[1]); if ((diskp=mmap(NULL, 512, PROT_READ | PROT_WRITE, 0, fd, 0)) == MAP_FAILED) err(EX_IOERR, Can't mmap() file); printf(%c\n,* (char *)diskp); * (char *)diskp = '!'; if (msync(diskp, 0, MS_SYNC) || close(fd)) err(EX_IOERR, Error closing file); All proceeds w/o errors with the sample (2 bytes long) file, printf() shows actual first byte of my file. But modification doesn't get written back to the disk, file contents are unchanged after execution of my code. I'm sure I'm overlooking something very basic and stupid, but can't find what exactly. Sincerely, Dmitry -- Atlantis ISP, System Administrator e-mail: [EMAIL PROTECTED] nic-hdl: LYNX-RIPE ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Simple question about mmap() system call
On Fri, 23 Jun 2006 17:47:57 +0300 (EEST) Dmitry Pryanishnikov [EMAIL PROTECTED] wrote: Hello! I'm writing an utility that should examine some bytes of a large file and modify them - that't all. I've decided to mmap() the file: void *diskp; if ((fd=open(argv[1], O_RDWR)) == -1) err(EX_NOINPUT, Can't open %s for readind and writing, argv[1]); if ((diskp=mmap(NULL, 512, PROT_READ | PROT_WRITE, 0, fd, 0)) == MAP_FAILED) err(EX_IOERR, Can't mmap() file); printf(%c\n,* (char *)diskp); * (char *)diskp = '!'; if (msync(diskp, 0, MS_SYNC) || close(fd)) err(EX_IOERR, Error closing file); All proceeds w/o errors with the sample (2 bytes long) file, printf() shows actual first byte of my file. But modification doesn't get written back to the disk, file contents are unchanged after execution of my code. I'm sure I'm overlooking something very basic and stupid, but can't find what exactly. munmap(2)? -- Stanislav Sedov MBSD labs, Inc. [EMAIL PROTECTED] Россия, Москва http://mbsd.msk.ru If the facts don't fit the theory, change the facts. -- A. Einstein PGP fingerprint: F21E D6CC 5626 9609 6CE2 A385 2BF5 5993 EB26 9581 pgp0H2fDxyodw.pgp Description: PGP signature
Buses, devices and modules
Hi all, I have a question regarding probe and attach routines. I've got 2 modules: for bus (mybus.ko) and for device (mydev.ko) on that bus. mydev.ko has MODULE_DEPEND(mydev, mybus, 1, 1, 1); When kldloading mydev.ko, mybus.ko is loading automatically, then it founds its device, attaches properly, and mydev.ko after that founds its own device, and attaches to the bus, and all works fine. When I do 'echo mydev_load=YES /boot/loader.conf', loader loads both mydev.ko and mybus.ko, and mybus.ko attaches properly, but mydev.ko don't attach to anything. I suppose this happens because mydev happens to probe for devices before mybus creates them, and fails to find anything to attach to. I'm looking for way to trigger the mydev driver to re-probe after the mybus driver actually creates the devices, or may be I'm missing something simple (like priority of module probing)? ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Simple question about mmap() system call
In the last episode (Jun 23), Dmitry Pryanishnikov said: I'm writing an utility that should examine some bytes of a large file and modify them - that't all. I've decided to mmap() the file: void *diskp; if ((fd=open(argv[1], O_RDWR)) == -1) err(EX_NOINPUT, Can't open %s for readind and writing, argv[1]); if ((diskp=mmap(NULL, 512, PROT_READ | PROT_WRITE, 0, fd, 0)) == MAP_FAILED) err(EX_IOERR, Can't mmap() file); Try adding MAP_SHARED. mmap defaults to private mappings, which means you changes don't get synched back to disk. I wonder how many programs would break if the mmap syscall returned an error if neither MAP_PRIVATE or MAP_SHARED were set... -- Dan Nelson [EMAIL PROTECTED] ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Simple question about mmap() system call
Hello! On Fri, 23 Jun 2006, Konstantin Belousov wrote: if ((diskp=mmap(NULL, 512, PROT_READ | PROT_WRITE, 0, fd, 0)) == MAP_FAILED) err(EX_IOERR, Can't mmap() file); shows actual first byte of my file. But modification doesn't get written back to the disk, file contents are unchanged after execution of my code. I'm sure I'm overlooking something very basic and stupid, but can't find what exactly. You forgot MAP_SHARED. Thank you! Yes, that's it. I would say that it's description MAP_SHAREDModifications are shared. is rather terse and doesn't explicitly says that w/o it data will not be stored back to the mmapped object. One can only deduce this behaviour, and deduction sometimes fails at the end of working day ;) Sincerely, Dmitry -- Atlantis ISP, System Administrator e-mail: [EMAIL PROTECTED] nic-hdl: LYNX-RIPE ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Virtual device driver programming
Robert Watson wrote: On Fri, 23 Jun 2006, Johnny Choque wrote: I'm interested in programming a virtual network device driver -Linux concept- on a FreeBSD box. The idea behind of this sort of interface is the following: From the kernel's point of view, a network interface is a software object that can process outgoing packets, and the actual transmission mechanism remains hidden inside the interface driver. Even though most interfaces are associated to physical devices (or, for the loopback interface, to a software-only data loop), it is possible to design network interface drivers that rely on other interfaces to perform actual packet transmission. The idea of a ``virtual'' interface can be useful to implement special-purpose processing on data packets while avoiding to hack with the network subsystem of the kernel. I know that is not too complicated to program this sort of functionality in linux but I would like to do it over freebsd, has anybody some idea on how could I start doing it? I've been searching in the freebsd handbook but I haven't found anything really relevant. tap(4) and tun(4) describe pseudo-devices you can use to instantiate ethernet and tunnel interfaces from user space. Programs attach to pseudo-devices, and using read/write operations on the pseudo-device, can receive and generate packets on the network interface. In kernel, the ifnet(9) API is used to implement network interfaces -- nothing in the API requires that the under-side of a network interface be hardware. In fact, a great many network types without underlying hardware have been implemented, including the loopback interface, encapsulation interfaces, and the tap/tun interface drivers. there is also gif(4) as well and if that is not enough the netgraph system gives you a toolkit to make your own kernel based virtual network interfaces. (ng_iface, ng_ether, ng_eiface (and others)) Robert N M Watson Computer Laboratory University of Cambridge ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to [EMAIL PROTECTED] ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to [EMAIL PROTECTED]
mmap() vs. character special file
Hello! mmap(2) manpage has the following title: NAME mmap -- allocate memory, or map files or devices into memory I'm curious about mmap()ing devices (particularly, HDD slices). The manpage mentions character special files only once: [EINVAL] MAP_ANON has not been specified and fd did not refer- ence a regular or character special file. So it looks like one should be able to mmap() a character special file (w/o MAP_ANON, of course). However, if I try to issue: if ((fd=open(argv[1], O_RDWR)) == -1) err(EX_NOINPUT, Can't open %s for readind and writing, argv[1]); if ((diskp=mmap(NULL, label_offset + sizeof (struct disklabel), PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0)) == MAP_FAILED) err(EX_IOERR, Can't mmap() file); against /dev/adXsY (adXsY and adXsYZ are _not_ mounted), I'm getting EINVAL from mmap(). Is mmap()ping a disk slice/partition impossible by design, or it just isn't implemented yet? IMHO manpage doesn't reply to this question. Sincerely, Dmitry -- Atlantis ISP, System Administrator e-mail: [EMAIL PROTECTED] nic-hdl: LYNX-RIPE ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: mmap() vs. character special file
On Sat, 24 Jun 2006 01:27:05 +0300 (EEST) Dmitry Pryanishnikov [EMAIL PROTECTED] wrote: Hello! mmap(2) manpage has the following title: NAME mmap -- allocate memory, or map files or devices into memory I'm curious about mmap()ing devices (particularly, HDD slices). The manpage mentions character special files only once: [EINVAL] MAP_ANON has not been specified and fd did not refer- ence a regular or character special file. So it looks like one should be able to mmap() a character special file (w/o MAP_ANON, of course). However, if I try to issue: if ((fd=open(argv[1], O_RDWR)) == -1) err(EX_NOINPUT, Can't open %s for readind and writing, argv[1]); if ((diskp=mmap(NULL, label_offset + sizeof (struct disklabel), PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0)) == MAP_FAILED) err(EX_IOERR, Can't mmap() file); against /dev/adXsY (adXsY and adXsYZ are _not_ mounted), I'm getting EINVAL from mmap(). Is mmap()ping a disk slice/partition impossible by design, or it just isn't implemented yet? IMHO manpage doesn't reply to this question. You cannot mmap ata devices (as well as scsi ones), since mmap functions was not implemented. Actually, only few devices have such support (e.g. drm, bktr). mmap requires device to return pointer to contigues memory range - drm, for example, returns pointer to physical memory of the graphical adapter. Implementing mmap for disk devices involves complicated VM intercations, since you cannot simply return pointer to physical memory. Actually, implementing mmap, IMHO, doesn't worth efforts to do this. -- Stanislav Sedov MBSD labs, Inc. [EMAIL PROTECTED] Россия, Москва http://mbsd.msk.ru If the facts don't fit the theory, change the facts. -- A. Einstein PGP fingerprint: F21E D6CC 5626 9609 6CE2 A385 2BF5 5993 EB26 9581 pgppOzxyCuGx9.pgp Description: PGP signature
