- Queries
- All Stories
- Search
- Advanced Search
- Transactions
- Transaction Logs
Feed Advanced Search
Advanced Search
Advanced Search
Oct 18 2019
Oct 18 2019
Make rt_getifa_fib() static.
Oct 17 2019
Oct 17 2019
glebius committed rS353695: Revert two parts of r353292 that enter epoch when processing vlan capabilities..
Revert two parts of r353292 that enter epoch when processing vlan capabilities.
glebius committed rS353687: Quickly fix up r353683: enter the epoch before calling into netisr_dispatch()..
Quickly fix up r353683: enter the epoch before calling into netisr_dispatch().
igmp_v1v2_queue_report() doesn't require epoch.
Oct 16 2019
Oct 16 2019
do_link_state_change() is executed in taskqueue context and in
Oct 15 2019
Oct 15 2019
When assertion for a thread not being in an epoch fails also print all
Re-do the patch a bit. Instead of new macro, just add epoch_trace_list()
to the three critical places where we must not be in an epoch.
glebius committed rS353566: Remove pfctlinput2(). It came from KAME and had never ever been in use..
Remove pfctlinput2(). It came from KAME and had never ever been in use.
Oct 14 2019
Oct 14 2019
glebius committed rS353522: if_delmulti() is never called without ifp argument, assert this instead.
if_delmulti() is never called without ifp argument, assert this instead
Convert to if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Convert if_foreach_llmaddr() KPI.
Convert to if_foreach_llmaddr() KPI.
Use epoch(9) directly instead of obsoleted KPI.
ipfw(4) rule matching always happens in network epoch.
in6ifa_llaonifp() is never called from fast path, so do not require
glebius committed rS353485: Since EPOCH_TRACE had been moved to opt_global.h, we don't need to waste.
Since EPOCH_TRACE had been moved to opt_global.h, we don't need to waste
glebius committed rS353484: Revert r353313. It is not needed with r353357 and is actually incorrect..
Revert r353313. It is not needed with r353357 and is actually incorrect.
Oct 13 2019
Oct 13 2019
vlan_config() isn't always called in epoch context.
Don't cover in6_ifattach() with network epoch, as it may call into
Oct 10 2019
Oct 10 2019
Don't use if_maddr_rlock() in 802.11, use epoch(9) directly instead.
Don't use if_maddr_rlock() in sppp(4), use epoch(9) directly instead.
glebius committed rS353425: Don't use if_maddr_rlock() in tuntap(4), use epoch(9) directly instead..
Don't use if_maddr_rlock() in tuntap(4), use epoch(9) directly instead.
glebius committed rS353424: Interface output method must be executed in network epoch, so if_addr_rlock().
Interface output method must be executed in network epoch, so if_addr_rlock()
glebius committed rS353423: Don't use if_maddr_rlock() in ng_eiface(4), use epoch(9) directly instead..
Don't use if_maddr_rlock() in ng_eiface(4), use epoch(9) directly instead.
The divert(4) module must always be running in network epoch, thus
glebius committed rS353421: Don't use if_maddr_rlock() in ng_ether(4), use epoch(9) directly instead..
Don't use if_maddr_rlock() in ng_ether(4), use epoch(9) directly instead.
Add two extra functions that basically give count of addresses
Provide new KPI for network drivers to access lists of interface
Oct 9 2019
Oct 9 2019
glebius added a comment to D21943: Correct and final KPI to traverse through interface address lists for drivers.Removal of not stack allocated epoch_tracker..
The plan is to commit first series of commits (see my github link), then do drivers in small batches, asking for testing and resorting to committing untested after timeout, if no volunteers found. When all drivers are done, finalize with the last two commits. Looks like we are all positive, so I'm going to proceed.
glebius updated the diff for D21943: Correct and final KPI to traverse through interface address lists for drivers.Removal of not stack allocated epoch_tracker..
Address reviewers comments.
Revert most of the multicast changes from r353292. This needs a more
ip6_output() has a complex set of gotos, and some can jump out of
Cleanup unneeded includes that crept in with r353292.
ifnet_byindex_ref() requires network epoch.
Enter network epoch in domain callouts.
Revert changes to rip6_bind() from r353292. This function is always
Oct 8 2019
Oct 8 2019
glebius committed rS353314: Remove epoch assertion from if_setlladdr(). Originally this function was.
Remove epoch assertion from if_setlladdr(). Originally this function was
glebius retitled D21943: Correct and final KPI to traverse through interface address lists for drivers.Removal of not stack allocated epoch_tracker. from Correct and final KBI to traverse through interface address lists for drivers.Removal of not stack allocated epoch_tracker. to Correct and final KPI to traverse through interface address lists for drivers.Removal of not stack allocated epoch_tracker..
glebius retitled D21943: Correct and final KPI to traverse through interface address lists for drivers.Removal of not stack allocated epoch_tracker. from Correct and final ABI to traverse through interface address lists for drivers.
Removal of not stack allocated epoch_tracker. to Correct and final KBI to traverse through interface address lists for drivers.Removal of not stack allocated epoch_tracker..
glebius committed rS353313: Quickly plug another regression from r353292. Again, multicast locking needs.
Quickly plug another regression from r353292. Again, multicast locking needs
glebius committed rS353312: In DIAGNOSTIC block of if_delmulti_ifma_flags() enter the network epoch..
In DIAGNOSTIC block of if_delmulti_ifma_flags() enter the network epoch.
Oct 7 2019
Oct 7 2019
Oct 3 2019
Oct 3 2019
Fix build failure from r353026. Somehow module build allowed this.
Protect access to seq->xwin[] with the seq mutex.
- Remove the compile time limit for number of links a ng_bridge node
Sep 30 2019
Sep 30 2019
In D19111#477072, @hselasky wrote:Hi,
I see wrappers are missing for WLAN:
net80211/ieee80211_input.c: ifp->if_input(ifp, m);
net80211/ieee80211_hostap.c: ifp->if_input(ifp, m);??
Sep 29 2019
Sep 29 2019
glebius added inline comments to D21179: Add header definition for RFC4340, Datagram Congestion Control Protocol.
Sep 28 2019
Sep 28 2019
@gallatin yes, I already got a branch that does this. Created it for Olivier Cochard's benchmarking of ipfw. https://github.com/glebius/FreeBSD/tree/net_epoch+IFF_KNOWSEPOCH+pfil
Sep 27 2019
Sep 27 2019
@bz That's fine, no hurry. I'm not sure that enter/exit points are 100% matching for VNETs. For example there is nothing wrong for a packet to travel between VNETs within one epoch.
@hselasky Good catch, thanks.
glebius added a reviewer for D19111: Summary: widen net_epoch coverage up to all packet processing: bz.
- Enter the net_epoch in uether_rxflush(). While here, rewrite the whole
glebius added inline comments to D19111: Summary: widen net_epoch coverage up to all packet processing.
Sep 26 2019
Sep 26 2019
I'd like to go forward with this change. It will bring a number of new epoch recursions, that don't exist today, but I want to resolve them separately, one per subsystem, otherwise the changeset will get bigger and more messy. I need your review, guys. :)
glebius added reviewers for D19111: Summary: widen net_epoch coverage up to all packet processing: • hselasky, kp, gallatin.
- WIP: compiles, boots in VM
- Enter epoch in DAD timer.
- Revert entering epoch for setsockopt. This needs to be fine grained.
- - In pfi_initialize_vnet() rely on IFNET_RLOCK() which is sx(9) for
- Revert if_purgeaddrs() change. This function is called in control path.
- Enter epoch for duration of ip6_output() just as we do for IPv4.
- Enter epoch in route_output().
- Revert this branch changes to in6_joingroup_locked(). Function can
- Another ND6 timer.
- Enter epoch before calling nd6_prefix_onlink() and nd6_lookup().
- pfi_kif_update() now requires epoch.
- Resolve rebase glitch.
- Revert back defrouter_select_fib().
- Fix few lock order issues.
- Revert changes to ifa_maintain_loopback_route().
- Merge remote-tracking branch 'FreeBSD/master' into net_epoch
- Merge branch 'epoch-trace' into net_epoch
- Merge branch 'epoch-trace' into net_epoch
- Merge branch 'master' into net_epoch
- Catch up on c6f2373c081e1431319aefa29218d59c1d01ff30.
- Fix mismerge.
- Merge remote-tracking branch 'FreeBSD/master' into net_epoch
- Restore whitespace to what we have in head.
- Merge remote-tracking branch 'FreeBSD/master' into net_epoch
Move EPOCH_TRACE to opt_global.h, so that any external modules that
Sep 25 2019
Sep 25 2019
style(9): remove extraneous empty lines
Add debugging facility EPOCH_TRACE that checks that epochs entered are
Enhance the 'ps' command so that it prints a line per proc and a line
Sep 24 2019
Sep 24 2019
- Merge remote-tracking branch 'FreeBSD/master' into epoch-trace
- More from Hans's review.
Sep 13 2019
Sep 13 2019
Drivers may pass runt packets to filter. This is okay.
- Improve per Hans's suggestions.
Sep 11 2019
Sep 11 2019
Sep 10 2019
Sep 10 2019
Fix 'calendar -a' in several ways.
glebius committed rS352115: Remove pointless playing with LC_TIME, which should have been done in r205821..
Remove pointless playing with LC_TIME, which should have been done in r205821.
Aug 30 2019
Aug 30 2019
Use mbuf queue instead of ifqueue in USB network drivers.
Allow mbuf queues to be unlimited.
Aug 16 2019
Aug 16 2019