* Moved eth_* to ne2k_* for future expansion @ 2017-03-04 16:46 Jody Bruchon 2017-03-04 17:48 ` Marc-François LUCCA-DANIAU 0 siblings, 1 reply; 5+ messages in thread From: Jody Bruchon @ 2017-03-04 16:46 UTC (permalink / raw To: ELKS I generalized some network stuff in case other network adapters are supported in the future. Make sure I didn't break anything, -Jody ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Moved eth_* to ne2k_* for future expansion 2017-03-04 16:46 Moved eth_* to ne2k_* for future expansion Jody Bruchon @ 2017-03-04 17:48 ` Marc-François LUCCA-DANIAU 2017-03-05 20:37 ` Georg Potthast 0 siblings, 1 reply; 5+ messages in thread From: Marc-François LUCCA-DANIAU @ 2017-03-04 17:48 UTC (permalink / raw To: ELKS OK for the changes that seem logical to me, except: * elks/arch/i86/drivers/net/ne2k-main.c : I actually planned to rename this file eth-main.c, because the pattern of any Ethernet driver is the same, and will implement the same "fops". Only the call to ne2k_* functions are specific. So we should keep the eth_* functions inside this module, that are generic. * elks/arch/i86/drivers/net/ne2k-test.c : renaming the variables related to Ethernet protocol and not linked to the NE2K has no sense : ne2k_from and ne2k_to, for example, are MAC adresses, and do not depend on the underlying Ethernet device. So we should rename them back to eth_from and eth_to. * elkscmd/test/socket/echo/.gitignore : I believe that Georg removed *_un and *_in programs, and replaced them by more generic programs ? If confirmed, remove *_un and *_in. * qemu.sh : "system" keyword means that we use the "QEMU system emulator". Not the "QEMU program emulator", that is not the same. So keep that accurate comment. MFLD 2017-03-04 17:46 GMT+01:00 Jody Bruchon <jody@jodybruchon.com>: > I generalized some network stuff in case other network adapters are > supported in the future. Make sure I didn't break anything, > > -Jody > -- > To unsubscribe from this list: send the line "unsubscribe linux-8086" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Moved eth_* to ne2k_* for future expansion 2017-03-04 17:48 ` Marc-François LUCCA-DANIAU @ 2017-03-05 20:37 ` Georg Potthast 2017-03-05 21:30 ` Marc-F. LUCCA-DANIAU 0 siblings, 1 reply; 5+ messages in thread From: Georg Potthast @ 2017-03-05 20:37 UTC (permalink / raw To: Marc-François LUCCA-DANIAU, ELKS When I look at the drivers in the char directory, most implement their own version of a fops structure and have an init function for the kernel. So I would expect that an RTL NIC driver would have the same structure as the NE2K driver, again with a fops structure and an init function. I would rather use this template than branch to various specific functions from a generic ethernet driver. This mainly because that has been the pattern used by ELKS so far and the developers may be more familiar with it. The *_un and *_in files have been deleted in the "elkscmd/test/socket/echo" directory and are replaced with echoserver and echoclient files that will use unix domain sockets if "-u" is passed on the command line. Georg -----Ursprüngliche Nachricht----- From: Marc-François LUCCA-DANIAU Sent: Saturday, March 4, 2017 6:48 PM To: ELKS Subject: Re: Moved eth_* to ne2k_* for future expansion OK for the changes that seem logical to me, except: * elks/arch/i86/drivers/net/ne2k-main.c : I actually planned to rename this file eth-main.c, because the pattern of any Ethernet driver is the same, and will implement the same "fops". Only the call to ne2k_* functions are specific. So we should keep the eth_* functions inside this module, that are generic. * elks/arch/i86/drivers/net/ne2k-test.c : renaming the variables related to Ethernet protocol and not linked to the NE2K has no sense : ne2k_from and ne2k_to, for example, are MAC adresses, and do not depend on the underlying Ethernet device. So we should rename them back to eth_from and eth_to. * elkscmd/test/socket/echo/.gitignore : I believe that Georg removed *_un and *_in programs, and replaced them by more generic programs ? If confirmed, remove *_un and *_in. * qemu.sh : "system" keyword means that we use the "QEMU system emulator". Not the "QEMU program emulator", that is not the same. So keep that accurate comment. MFLD 2017-03-04 17:46 GMT+01:00 Jody Bruchon <jody@jodybruchon.com>: > I generalized some network stuff in case other network adapters are > supported in the future. Make sure I didn't break anything, > > -Jody > -- > To unsubscribe from this list: send the line "unsubscribe linux-8086" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe linux-8086" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Moved eth_* to ne2k_* for future expansion 2017-03-05 20:37 ` Georg Potthast @ 2017-03-05 21:30 ` Marc-F. LUCCA-DANIAU 2017-03-06 12:28 ` Georg Potthast 2 0 siblings, 1 reply; 5+ messages in thread From: Marc-F. LUCCA-DANIAU @ 2017-03-05 21:30 UTC (permalink / raw To: ELKS Sorry Georg, but I won't spend any time in supporting more Ethernet devices, because I think that: 1) NE2K is the only ISA device provided by QEMU, other need PCI support, and ELKS has no PCI support. I don't think adding PCI support would be in the scope of this project, because PCI came with the 32 bits generation. 2) My goal is to run ELKS on my real SBC, that has a NE2K clone chip. I am therefore not interested in any other Ethernet chip. 3) ELKS is missing the network driver framework, so a discussion for a generic driver is biased by this limitation. IMHO, the best driver framework is one generic Ethernet driver, and specialized packet drivers, with a standard "packet ops" interface for clean branching. The template you are dealing with should be not at Ethernet driver level, but at packet driver level. But again, back to 1) and 2), I see no need to develop that framework. MFLD Le 05/03/2017 à 21:37, Georg Potthast a écrit : > When I look at the drivers in the char directory, most implement their > own version of a fops structure and have an init function for the > kernel. So I would expect that an RTL NIC driver would have the same > structure as the NE2K driver, again with a fops structure and an init > function. I would rather use this template than branch to various > specific functions from a generic ethernet driver. This mainly because > that has been the pattern used by ELKS so far and the developers may > be more familiar with it. > > The *_un and *_in files have been deleted in the > "elkscmd/test/socket/echo" directory and are replaced with echoserver > and echoclient files that will use unix domain sockets if "-u" is > passed on the command line. > > Georg > > -----Ursprüngliche Nachricht----- From: Marc-François LUCCA-DANIAU > Sent: Saturday, March 4, 2017 6:48 PM > To: ELKS > Subject: Re: Moved eth_* to ne2k_* for future expansion > > OK for the changes that seem logical to me, except: > > * elks/arch/i86/drivers/net/ne2k-main.c : I actually planned to rename > this file eth-main.c, because the pattern of any Ethernet driver is > the same, and will implement the same "fops". Only the call to ne2k_* > functions are specific. So we should keep the eth_* functions inside > this module, that are generic. > > * elks/arch/i86/drivers/net/ne2k-test.c : renaming the variables > related to Ethernet protocol and not linked to the NE2K has no sense : > ne2k_from and ne2k_to, for example, are MAC adresses, and do not > depend on the underlying Ethernet device. So we should rename them > back to eth_from and eth_to. > > * elkscmd/test/socket/echo/.gitignore : I believe that Georg removed > *_un and *_in programs, and replaced them by more generic programs ? > If confirmed, remove *_un and *_in. > > * qemu.sh : "system" keyword means that we use the "QEMU system > emulator". Not the "QEMU program emulator", that is not the same. So > keep that accurate comment. > > MFLD > > > > > 2017-03-04 17:46 GMT+01:00 Jody Bruchon <jody@jodybruchon.com>: >> I generalized some network stuff in case other network adapters are >> supported in the future. Make sure I didn't break anything, >> >> -Jody >> -- >> To unsubscribe from this list: send the line "unsubscribe linux-8086" in >> the body of a message to majordomo@vger.kernel.org >> More majordomo info at http://vger.kernel.org/majordomo-info.html > -- > To unsubscribe from this list: send the line "unsubscribe linux-8086" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Moved eth_* to ne2k_* for future expansion 2017-03-05 21:30 ` Marc-F. LUCCA-DANIAU @ 2017-03-06 12:28 ` Georg Potthast 2 0 siblings, 0 replies; 5+ messages in thread From: Georg Potthast 2 @ 2017-03-06 12:28 UTC (permalink / raw To: ELKS, Marc-F. LUCCA-DANIAU Marc-F., maybe someone else will find a way to support PCI using 16bit code in the future and will want to add an additional driver. So I would use a code structure which allows for additional NIC drivers. I know supporting wireless is difficult but it would be an interesting addition. Georg > "Marc-F. LUCCA-DANIAU" <mfld.fr@gmail.com> hat am 5. März 2017 um 22:30 geschrieben: > > > Sorry Georg, but I won't spend any time in supporting more Ethernet > devices, because I think that: > > 1) NE2K is the only ISA device provided by QEMU, other need PCI support, > and ELKS has no PCI support. I don't think adding PCI support would be > in the scope of this project, because PCI came with the 32 bits generation. > > 2) My goal is to run ELKS on my real SBC, that has a NE2K clone chip. I > am therefore not interested in any other Ethernet chip. > > 3) ELKS is missing the network driver framework, so a discussion for a > generic driver is biased by this limitation. IMHO, the best driver > framework is one generic Ethernet driver, and specialized packet > drivers, with a standard "packet ops" interface for clean branching. The > template you are dealing with should be not at Ethernet driver level, > but at packet driver level. But again, back to 1) and 2), I see no need > to develop that framework. > > MFLD > > > Le 05/03/2017 à 21:37, Georg Potthast a écrit : > > When I look at the drivers in the char directory, most implement their > > own version of a fops structure and have an init function for the > > kernel. So I would expect that an RTL NIC driver would have the same > > structure as the NE2K driver, again with a fops structure and an init > > function. I would rather use this template than branch to various > > specific functions from a generic ethernet driver. This mainly because > > that has been the pattern used by ELKS so far and the developers may > > be more familiar with it. > > > > The *_un and *_in files have been deleted in the > > "elkscmd/test/socket/echo" directory and are replaced with echoserver > > and echoclient files that will use unix domain sockets if "-u" is > > passed on the command line. > > > > Georg > > > > -----Ursprüngliche Nachricht----- From: Marc-François LUCCA-DANIAU > > Sent: Saturday, March 4, 2017 6:48 PM > > To: ELKS > > Subject: Re: Moved eth_* to ne2k_* for future expansion > > > > OK for the changes that seem logical to me, except: > > > > * elks/arch/i86/drivers/net/ne2k-main.c : I actually planned to rename > > this file eth-main.c, because the pattern of any Ethernet driver is > > the same, and will implement the same "fops". Only the call to ne2k_* > > functions are specific. So we should keep the eth_* functions inside > > this module, that are generic. > > > > * elks/arch/i86/drivers/net/ne2k-test.c : renaming the variables > > related to Ethernet protocol and not linked to the NE2K has no sense : > > ne2k_from and ne2k_to, for example, are MAC adresses, and do not > > depend on the underlying Ethernet device. So we should rename them > > back to eth_from and eth_to. > > > > * elkscmd/test/socket/echo/.gitignore : I believe that Georg removed > > *_un and *_in programs, and replaced them by more generic programs ? > > If confirmed, remove *_un and *_in. > > > > * qemu.sh : "system" keyword means that we use the "QEMU system > > emulator". Not the "QEMU program emulator", that is not the same. So > > keep that accurate comment. > > > > MFLD > > > > > > > > > > 2017-03-04 17:46 GMT+01:00 Jody Bruchon <jody@jodybruchon.com>: > >> I generalized some network stuff in case other network adapters are > >> supported in the future. Make sure I didn't break anything, > >> > >> -Jody > >> -- > >> To unsubscribe from this list: send the line "unsubscribe linux-8086" in > >> the body of a message to majordomo@vger.kernel.org > >> More majordomo info at http://vger.kernel.org/majordomo-info.html > > -- > > To unsubscribe from this list: send the line "unsubscribe linux-8086" in > > the body of a message to majordomo@vger.kernel.org > > More majordomo info at http://vger.kernel.org/majordomo-info.html > > > > -- > To unsubscribe from this list: send the line "unsubscribe linux-8086" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html ^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2017-03-06 12:28 UTC | newest] Thread overview: 5+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2017-03-04 16:46 Moved eth_* to ne2k_* for future expansion Jody Bruchon 2017-03-04 17:48 ` Marc-François LUCCA-DANIAU 2017-03-05 20:37 ` Georg Potthast 2017-03-05 21:30 ` Marc-F. LUCCA-DANIAU 2017-03-06 12:28 ` Georg Potthast 2
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).