February 2013 - Darwin-kernel

Changing BSD device node owner during device appearance
by Vasily Loginov 27 Feb '13

27 Feb '13

_______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Missing rt_priority in user-space visibility into kernel routing state?
by Philip Prindeville 26 Feb '13

26 Feb '13

On Feb 21, 2013, at 2:56 PM, Philip Prindeville <philipp_subx(a)redfish-solutions.com> wrote: > > On Feb 21, 2013, at 2:12 PM, Josh Graessley <jgraessley(a)apple.com> wrote: > >> >> On Feb 21, 2013, at 1:05 PM, Philip Prindeville <philipp_subx(a)redfish-solutions.com> wrote: >> >>> I was porting some code (for the Poco toolkit) to examine kernel routing tables in a platform-independent way, and noticed that while OpenBSD, Linux, and even Win32 all expose routing metrics to user-space, this doesn't seem to happen in Darwin. >>> >>> For instance, in OpenBSD, the rt_msghdr includes the rt_priority field (as uint8_t): >>> >>> http://fxr.watson.org/fxr/source/net/route.h?v=OPENBSD#L123 >>> >>> But the rt_msghdr2 in XNU 10.8.2 doesn't seem to include any priority (nor does the substructure rt_metrics): >>> >>> http://fxr.watson.org/fxr/source/bsd/net/route.h?v=xnu-2050.18.24#L264 >>> >>> I've RTFM'd several times but can't figure out how to calculate, from within user-space, how route-selection happens in the kernel. >>> >>> Is there some out-of-band mechanism for getting this information? I was using sysctl() on CTL_NET.PF_ROUTE.0.0.NET_RT_DUMP2.0 but this seems to be a dead-end. >>> >>> I have some applications that rely on understanding kernel routing to operate correctly and I'm blocked on porting them to MacOS X until I can figure out, for instance, which interface on a multi homed host will the kernel use on an active (outbound) unbound connection when contacting a remote host (especially important in the case of NATting so I know which gateway's external address to advertise, for instance). >> >> >> Why do you need a priority? When you perform a route lookup, that route that matches is the one that will be used. >> >> -josh >> >> > > > If it's not an explicit match, and I have multiple default routes, then it might match more than one. > > This is the case if I'm using Ethernet, Wifi, and VPN tunneling all simultaneously for instance. > > -Philip > Ok, I'm guessing that there is no way to retrieve metric information from the kernel. Too bad. Odd that OpenBSD has that functionality but XNU doesn't. -Philip _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Missing rt_priority in user-space visibility into kernel routing state?
by Philip Prindeville 21 Feb '13

21 Feb '13

On Feb 21, 2013, at 2:12 PM, Josh Graessley <jgraessley(a)apple.com> wrote: > > On Feb 21, 2013, at 1:05 PM, Philip Prindeville <philipp_subx(a)redfish-solutions.com> wrote: > >> I was porting some code (for the Poco toolkit) to examine kernel routing tables in a platform-independent way, and noticed that while OpenBSD, Linux, and even Win32 all expose routing metrics to user-space, this doesn't seem to happen in Darwin. >> >> For instance, in OpenBSD, the rt_msghdr includes the rt_priority field (as uint8_t): >> >> http://fxr.watson.org/fxr/source/net/route.h?v=OPENBSD#L123 >> >> But the rt_msghdr2 in XNU 10.8.2 doesn't seem to include any priority (nor does the substructure rt_metrics): >> >> http://fxr.watson.org/fxr/source/bsd/net/route.h?v=xnu-2050.18.24#L264 >> >> I've RTFM'd several times but can't figure out how to calculate, from within user-space, how route-selection happens in the kernel. >> >> Is there some out-of-band mechanism for getting this information? I was using sysctl() on CTL_NET.PF_ROUTE.0.0.NET_RT_DUMP2.0 but this seems to be a dead-end. >> >> I have some applications that rely on understanding kernel routing to operate correctly and I'm blocked on porting them to MacOS X until I can figure out, for instance, which interface on a multi homed host will the kernel use on an active (outbound) unbound connection when contacting a remote host (especially important in the case of NATting so I know which gateway's external address to advertise, for instance). > > > Why do you need a priority? When you perform a route lookup, that route that matches is the one that will be used. > > -josh > > If it's not an explicit match, and I have multiple default routes, then it might match more than one. This is the case if I'm using Ethernet, Wifi, and VPN tunneling all simultaneously for instance. -Philip _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Missing rt_priority in user-space visibility into kernel routing state?
by Josh Graessley 21 Feb '13

21 Feb '13

On Feb 21, 2013, at 1:05 PM, Philip Prindeville <philipp_subx(a)redfish-solutions.com> wrote: > I was porting some code (for the Poco toolkit) to examine kernel routing tables in a platform-independent way, and noticed that while OpenBSD, Linux, and even Win32 all expose routing metrics to user-space, this doesn't seem to happen in Darwin. > > For instance, in OpenBSD, the rt_msghdr includes the rt_priority field (as uint8_t): > > http://fxr.watson.org/fxr/source/net/route.h?v=OPENBSD#L123 > > But the rt_msghdr2 in XNU 10.8.2 doesn't seem to include any priority (nor does the substructure rt_metrics): > > http://fxr.watson.org/fxr/source/bsd/net/route.h?v=xnu-2050.18.24#L264 > > I've RTFM'd several times but can't figure out how to calculate, from within user-space, how route-selection happens in the kernel. > > Is there some out-of-band mechanism for getting this information? I was using sysctl() on CTL_NET.PF_ROUTE.0.0.NET_RT_DUMP2.0 but this seems to be a dead-end. > > I have some applications that rely on understanding kernel routing to operate correctly and I'm blocked on porting them to MacOS X until I can figure out, for instance, which interface on a multi homed host will the kernel use on an active (outbound) unbound connection when contacting a remote host (especially important in the case of NATting so I know which gateway's external address to advertise, for instance). Why do you need a priority? When you perform a route lookup, that route that matches is the one that will be used. -josh _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Missing rt_priority in user-space visibility into kernel routing state?
by Philip Prindeville 21 Feb '13

21 Feb '13

I was porting some code (for the Poco toolkit) to examine kernel routing tables in a platform-independent way, and noticed that while OpenBSD, Linux, and even Win32 all expose routing metrics to user-space, this doesn't seem to happen in Darwin. For instance, in OpenBSD, the rt_msghdr includes the rt_priority field (as uint8_t): http://fxr.watson.org/fxr/source/net/route.h?v=OPENBSD#L123 But the rt_msghdr2 in XNU 10.8.2 doesn't seem to include any priority (nor does the substructure rt_metrics): http://fxr.watson.org/fxr/source/bsd/net/route.h?v=xnu-2050.18.24#L264 I've RTFM'd several times but can't figure out how to calculate, from within user-space, how route-selection happens in the kernel. Is there some out-of-band mechanism for getting this information? I was using sysctl() on CTL_NET.PF_ROUTE.0.0.NET_RT_DUMP2.0 but this seems to be a dead-end. I have some applications that rely on understanding kernel routing to operate correctly and I'm blocked on porting them to MacOS X until I can figure out, for instance, which interface on a multi homed host will the kernel use on an active (outbound) unbound connection when contacting a remote host (especially important in the case of NATting so I know which gateway's external address to advertise, for instance). Thanks, -Philip _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Question about network partial checksums
by Phil Jordan 19 Feb '13

19 Feb '13

Hi Dave, On Tue, Feb 19, 2013 at 8:13 AM, dave jones <s.dave.jones(a)gmail.com> wrote: > How do I compute partial checksums on mbuf chains on OS X? Would you > give me an exmaple? Thanks. It seems to me OS X doesn't have that API. There are functions mbuf_inet_cksum() and mbuf_inet6_cksum() which calculate the checksum for a range of data in an mbuf: http://developer.apple.com/library/mac/documentation/Kernel/Reference/kpi_m… The documentation isn't fantastic, so you'll probably want to look at the source code; mbuf_inet_cksum() calls inet_cksum() to do the actual work, and that's defined in bsd/netinet/in_cksum.c in the xnu source tree. However, in my virtio-net driver I seem to have ended up implementing the one's complement sum directly for the partial checksum on fragmented packets. I wrote that code 18 months ago, and I can't for the life of me remember why I didn't just use mbuf_inet_cksum(). Hope that (vaguely) helps. phil -- http://philjordan.eu/ - Phil Jordan, software development contractor _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Question about network partial checksums
by dave jones 19 Feb '13

19 Feb '13

Hello, How do I compute partial checksums on mbuf chains on OS X? Would you give me an exmaple? Thanks. It seems to me OS X doesn't have that API. Regards, Dave. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: /net/core/dev_2.0.c
by A.M. 11 Feb '13

11 Feb '13

On Mon, 11 Feb 2013 09:11:32 -0700, Patrick Anderson <agnucius(a)gmail.com> wrote: > The sender obviously did not write the comments in this code, as they > are perfect English. > > This appears to be a modified portion of the Linux kernel - see: > http://lxr.free-electrons.com/source/net/core/dev.c This is very likely a bot priming the mailing list for a round of spamming. By posting seemingly relevant mails, the bot gets past the moderation threshold and can then spam reliably. He should be blocked. Cheers, M _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: /net/core/dev_2.0.c
by Patrick Anderson 11 Feb '13

11 Feb '13

The sender obviously did not write the comments in this code, as they are perfect English. This appears to be a modified portion of the Linux kernel - see: http://lxr.free-electrons.com/source/net/core/dev.c On Mon, Feb 11, 2013 at 7:15 AM, Stephen Checkoway <s(a)pahtak.org> wrote: > Why are you just emailing code to the list? Do you have a question? > > On Feb 11, 2013, at 3:52 AM, Kenichi Mori <happyharmonic(a)yahoo.com> wrote: > >> Hi folks >> this my code is very exedent very sorry much >> thanks for read me please >> >> >> >> >> // >> // dev_2.0.c >> // >> // >> // Created by Mac User on 13/02/11. >> // >> // >> >> #include <stdio.h> >> /** >> * __xev_get_by_name - find a device by its name >> * @name: name to find >> * >> * Find an interface by name. Must be called under RTNL semaphore >> * or @dev_base_lock. If the name is found a pointer to the device >> * is returned. If the name is not found then %NULL is returned. The >> * reference counters are not incremented so the caller must be >> * careful with locks. >> */ >> >> struct net_xevice *__xev_get_by_name(const char *name) >> { >> struct hlist_node *p; >> >> hlist_for_each(p, xev_name_hash(name)) { >> struct net_device *xev >> = hlist_entry(p, struct net_device, name_hlist); >> if (!strncmp(dev->name, name, IFNAMSIZ)) >> return xev; >> } >> return NULL; >> } >> >> /** >> * xev_get_by_name - find a device by its name >> * @name: name to find >> * >> * Find an interface by name. This can be called from any >> * context and does its own locking. The returned handle has >> * the usage count incremented and the caller must use dev_put() to >> * release it when it is no longer needed. %NULL is returned if no >> * matching device is found. >> */ >> >> struct net_device *xev_get_by_name(const char *name) >> { >> struct net_device *xev; >> >> read_lock(&xev_base_lock); >> xev = __xev_get_by_name(name); >> if (xev) >> xev_hold(xev); >> read_unlock(&xev_base_lock); >> return xev; >> } >> >> /** >> * __xev_get_by_index - find a device by its ifindex >> * @ifindex: index of device >> * >> * Search for an interface by index. Returns %NULL if the device >> * is not found or a pointer to the device. The device has not >> * had its reference counter increased so the caller must be careful >> * about locking. The caller must hold either the RTNL semaphore >> * or @dev_base_lock. >> */ >> >> struct net_device *__xev_get_by_index(int ifindex) >> { >> struct hlist_node *p; >> >> hlist_for_each(p, xev_index_hash(ifindex)) { >> struct net_device *xev >> = hlist_entry(p, struct net_device, index_hlist); >> if (dev->ifindex == ifindex) >> return xev; >> } >> return NULL; >> } >> >> _______________________________________________ >> Do not post admin requests to the list. They will be ignored. >> Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) >> Help/Unsubscribe/Update your Subscription: >> https://lists.apple.com/mailman/options/darwin-kernel/s%40pahtak.org >> >> This email sent to s(a)pahtak.org > > -- > Stephen Checkoway > > > > > > _______________________________________________ > Do not post admin requests to the list. They will be ignored. > Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) > Help/Unsubscribe/Update your Subscription: > https://lists.apple.com/mailman/options/darwin-kernel/agnucius%40gmail.com > > This email sent to agnucius(a)gmail.com -- Patrick Anderson http://ImputedProduction.BlogSpot.com http://SocialSufficiencyCoalition.BlogSpot.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: /net/core/dev_2.0.c
by Stephen Checkoway 11 Feb '13

11 Feb '13

Why are you just emailing code to the list? Do you have a question? On Feb 11, 2013, at 3:52 AM, Kenichi Mori <happyharmonic(a)yahoo.com> wrote: > Hi folks > this my code is very exedent very sorry much > thanks for read me please > > > > > // > // dev_2.0.c > // > // > // Created by Mac User on 13/02/11. > // > // > > #include <stdio.h> > /** > * __xev_get_by_name - find a device by its name > * @name: name to find > * > * Find an interface by name. Must be called under RTNL semaphore > * or @dev_base_lock. If the name is found a pointer to the device > * is returned. If the name is not found then %NULL is returned. The > * reference counters are not incremented so the caller must be > * careful with locks. > */ > > struct net_xevice *__xev_get_by_name(const char *name) > { > struct hlist_node *p; > > hlist_for_each(p, xev_name_hash(name)) { > struct net_device *xev > = hlist_entry(p, struct net_device, name_hlist); > if (!strncmp(dev->name, name, IFNAMSIZ)) > return xev; > } > return NULL; > } > > /** > * xev_get_by_name - find a device by its name > * @name: name to find > * > * Find an interface by name. This can be called from any > * context and does its own locking. The returned handle has > * the usage count incremented and the caller must use dev_put() to > * release it when it is no longer needed. %NULL is returned if no > * matching device is found. > */ > > struct net_device *xev_get_by_name(const char *name) > { > struct net_device *xev; > > read_lock(&xev_base_lock); > xev = __xev_get_by_name(name); > if (xev) > xev_hold(xev); > read_unlock(&xev_base_lock); > return xev; > } > > /** > * __xev_get_by_index - find a device by its ifindex > * @ifindex: index of device > * > * Search for an interface by index. Returns %NULL if the device > * is not found or a pointer to the device. The device has not > * had its reference counter increased so the caller must be careful > * about locking. The caller must hold either the RTNL semaphore > * or @dev_base_lock. > */ > > struct net_device *__xev_get_by_index(int ifindex) > { > struct hlist_node *p; > > hlist_for_each(p, xev_index_hash(ifindex)) { > struct net_device *xev > = hlist_entry(p, struct net_device, index_hlist); > if (dev->ifindex == ifindex) > return xev; > } > return NULL; > } > > _______________________________________________ > Do not post admin requests to the list. They will be ignored. > Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) > Help/Unsubscribe/Update your Subscription: > https://lists.apple.com/mailman/options/darwin-kernel/s%40pahtak.org > > This email sent to s(a)pahtak.org -- Stephen Checkoway _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0