Page MenuHomeFreeBSD
People glebius

glebius (Gleb Smirnoff)
User

Projects (6)
View All

User Details

User Since
May 28 2014, 2:27 PM (608 w, 2 d)

Recent Activity
View All

Yesterday

glebius committed rGf3de667137e9: netinet6: free in6_ifextra with epoch_call(9).
Fri, Jan 23, 10:37 PM
glebius committed rGcba9f88105c2: netinet6: embed struct mld_ifsoftc into struct in6_ifextra.
Fri, Jan 23, 10:37 PM
glebius committed rGaa3bbc06e5f0: netinet6: embed struct scope6_id into struct in6_ifextra.
Fri, Jan 23, 10:37 PM
glebius committed rG576ee62dd2e5: netinet6: store ND context directly in struct in6_ifextra.
Fri, Jan 23, 10:37 PM
glebius committed rG25a11d9805fc: netinet6: use in6_ifmtu() instead of IN6_LINKMTU() macro.
Fri, Jan 23, 10:37 PM
glebius committed rG3182dc411788: netinet6: embed the counter(9) arrays in struct in6_ifextra.
Fri, Jan 23, 10:37 PM
glebius committed rG6857b3a04e59: netinet6: embed struct nd_ifinfo into struct in6_ndireq.
Fri, Jan 23, 10:37 PM
glebius committed rG37d11fde8f58: tests/kern: add test for SSL_sendfile().
Fri, Jan 23, 7:17 PM

Wed, Jan 21

glebius accepted D54763: Make sbin/ipfw on RELENG_14 compatible to ipfw kernel module on RELENG_15+.
Wed, Jan 21, 12:25 AM
glebius added a comment to D54763: Make sbin/ipfw on RELENG_14 compatible to ipfw kernel module on RELENG_15+.

To me the patch looks good enough. I can't approve the approach taken, that adds a future version tool to a stable/xx tree. This is up to releng as the main upgrade path gatekeeper and @jhb who requested the compatibility shim.

Wed, Jan 21, 12:25 AM

Tue, Jan 20

glebius accepted D54797: rpctls_impl.c: Use a direct cast to uintptr_t instead of __DECONST.
Tue, Jan 20, 5:43 PM
glebius added inline comments to D54763: Make sbin/ipfw on RELENG_14 compatible to ipfw kernel module on RELENG_15+.
Tue, Jan 20, 5:40 PM

Mon, Jan 19

glebius added inline comments to D54763: Make sbin/ipfw on RELENG_14 compatible to ipfw kernel module on RELENG_15+.
Mon, Jan 19, 6:45 PM
glebius accepted D54434: socket.2: Cross-reference protocol familiy man pages in section 4.

Looks great, thanks!

Mon, Jan 19, 6:31 PM · manpages

Sun, Jan 18

glebius committed rG76af334737b1: ipfw: fix !VIMAGE build.
Sun, Jan 18, 5:53 AM

Sat, Jan 17

glebius added inline comments to D54434: socket.2: Cross-reference protocol familiy man pages in section 4.
Sat, Jan 17, 6:01 PM · manpages
glebius accepted D53786: netlink(4): add snl(3) to See Also section.

Approved!

Sat, Jan 17, 5:52 PM
glebius committed rG35caa56a329d: tests/ipfw: add a test for ipfw(4) log rules that write to bpf(4).
Sat, Jan 17, 4:10 AM
glebius committed rG65b4bf7a6e0d: ipfw: refactor how we store bpf tap points.
Sat, Jan 17, 4:10 AM
glebius committed rG1a7b74d3125f: ipfw: in a vnet destructor use NET_EPOCH_WAIT().
Sat, Jan 17, 4:10 AM
glebius committed rGb0c13e7e2446: ipfw: remove locking workarounds in the table code.
Sat, Jan 17, 12:46 AM
glebius committed rG7eac31c83ee3: tests/ipfw: add a simple fuzzing test for ipfw tables.
Sat, Jan 17, 12:46 AM
glebius committed rGe3caa360d5d0: ipfw: make the upper half lock sleepable.
Sat, Jan 17, 12:46 AM
glebius closed D54580: ipfw: remove locking workarounds in the table code.
Sat, Jan 17, 12:46 AM
glebius closed D54579: tests/ipfw: add a simple fuzzing test for ipfw tables.
Sat, Jan 17, 12:46 AM
glebius closed D54535: ipfw: make the upper half lock sleepable.
Sat, Jan 17, 12:46 AM
glebius closed D54721: net: on interface detach purge all its routes before detaching protocols.
Sat, Jan 17, 12:46 AM
glebius committed rGf102a1834933: net: on interface detach purge all its routes before detaching protocols.
Sat, Jan 17, 12:45 AM

Fri, Jan 16

glebius added a comment to D54726: netinet6: embed struct scope6_id into struct in6_ifextra.

I have very little idea about that. I'm just working on making interfaces detach more safe and I want in6_ifextra to be a single piece of memory which would have delayed reclamation.

Fri, Jan 16, 7:00 PM

Thu, Jan 15

glebius committed rG51786cdeb07c: sockets: remove unused pr_sosend_notsupp(), pr_soreceive_notsupp().
Thu, Jan 15, 9:58 PM
glebius committed rG77f453e9f9ca: tests/if_ovpn: add missing cleanup to the multihome6 test.
Thu, Jan 15, 9:41 PM
glebius committed rGf9ae08a88924: linux: on vnet detach call clean_unrhdr(9) always.
Thu, Jan 15, 9:41 PM
glebius added a reviewer for D54725: netinet6: store ND context directly in struct in6_ifextra: madpilot.
Thu, Jan 15, 5:36 AM
glebius added a comment to D54636: rtadvd: add multi pref64 support.
In D54636#1250207, @pouria wrote:

@glebius should these minor changes to the manual be approved by the manpages group?

Thu, Jan 15, 3:37 AM
glebius requested review of D54728: netinet6: free in6_ifextra with epoch_call(9).
Thu, Jan 15, 3:25 AM
glebius requested review of D54727: netinet6: embed struct mld_ifsoftc into struct in6_ifextra.
Thu, Jan 15, 3:25 AM
glebius requested review of D54726: netinet6: embed struct scope6_id into struct in6_ifextra.
Thu, Jan 15, 3:25 AM
glebius requested review of D54725: netinet6: store ND context directly in struct in6_ifextra.
Thu, Jan 15, 3:25 AM
glebius requested review of D54724: netinet6: use in6_ifmtu() instead of IN6_LINKMTU() macro.
Thu, Jan 15, 3:24 AM
glebius requested review of D54721: net: on interface detach purge all its routes before detaching protocols.
Thu, Jan 15, 3:24 AM
glebius requested review of D54723: netinet6: embed the counter(9) arrays in struct in6_ifextra.
Thu, Jan 15, 3:24 AM
glebius requested review of D54722: netinet6: embed struct nd_ifinfo into struct in6_ndireq.
Thu, Jan 15, 3:24 AM
glebius committed rG008e5703e184: netinet6: garbage collect OSIOCGIFINFO_IN6.
Thu, Jan 15, 2:57 AM

Wed, Jan 14

glebius added a comment to D54718: tcpdump: Fix buffer overrun in ATM OAM printing .

Have you tried to submit this the tcpdump project itself? https://github.com/the-tcpdump-group/tcpdump

Wed, Jan 14, 6:29 PM
glebius accepted D54713: Fix sendfile over tmpfs.
Wed, Jan 14, 5:09 PM
glebius added inline comments to D54636: rtadvd: add multi pref64 support.
Wed, Jan 14, 5:07 PM
glebius committed rGc1376acb8a9d: ipfw: find_tentry method is defined for any table algorithm.
Wed, Jan 14, 1:01 AM
glebius committed rGe8348e2b35f3: enc: create an interface at SI_SUB_PROTO_IF stage.
Wed, Jan 14, 12:55 AM
glebius added inline comments to D54636: rtadvd: add multi pref64 support.
Wed, Jan 14, 12:50 AM
glebius added inline comments to D54535: ipfw: make the upper half lock sleepable.
Wed, Jan 14, 12:50 AM
glebius added a member for committers: pouria.
Wed, Jan 14, 12:49 AM

Mon, Jan 12

glebius accepted D54638: new committer (src): Pouria Mousavizadeh Tehrani (pouria).
Mon, Jan 12, 1:45 AM
glebius accepted D54637: committers-src: Add myself (pouria@) with glebius@ as mentor.
Mon, Jan 12, 1:44 AM

Sat, Jan 10

glebius committed rG509a185dd940: net80211: fix bpf tap leak on wlan(4) detach.
Sat, Jan 10, 6:57 PM

Thu, Jan 8

glebius added inline comments to D54607: ipfilter: Interface name must not extend beyond end of buffer.
Thu, Jan 8, 9:15 PM
glebius added a comment to D54605: iflib: remove convoluted custom zeroing code.

I would suggest to go deeper and call memset() directly. These empty proxy functions do not improve readability.

Thu, Jan 8, 6:27 PM
glebius accepted D54603: ip_mroute: Convert to using a regular mutex.
Thu, Jan 8, 5:38 PM
glebius retitled D54581: netinet: Drop zero-length mbufs in rip_send() to prevent kernel panic from netinet: Drop zero-length mbufs in ip_output to prevent kernel panic
Thu, Jan 8, 12:47 AM

Wed, Jan 7

glebius requested changes to D54581: netinet: Drop zero-length mbufs in rip_send() to prevent kernel panic.

The description mentions ip_output(), but in reality the patch is against rip_send(), which is send(2) method for a PF_INET/SOCK_RAW. Looks like with a specially crafted send(2) we can achieve such a bogus mbuf. Instead of putting the bandaid let's find out how this mbuf was produced and fix the bug. Note that mbufs are formed outside of protocol implementation, so potentially the problem discovered may apply to other protocols beyond PF_INET/SOCK_RAW.

Wed, Jan 7, 7:17 PM
glebius requested review of D54580: ipfw: remove locking workarounds in the table code.
Wed, Jan 7, 5:41 PM
glebius requested review of D54579: tests/ipfw: add a simple fuzzing test for ipfw tables.
Wed, Jan 7, 5:38 PM
glebius updated the diff for D54535: ipfw: make the upper half lock sleepable.
The only sleepable context where the lock was acquired was dyn_tick(). The
comment said it is done to prevent parallel execution of
dyn_expire_states().  However, there is proper internal locking in there
and function should be safe to execute in parallel.  The real problem is
dyn_expire_states() called via userland to race with dyn_grow_hashtable()
called via dyn_tick().  Protect against this condition with the main chain
lock.
Wed, Jan 7, 5:33 PM

Tue, Jan 6

glebius updated the diff for D54535: ipfw: make the upper half lock sleepable.
  • Fix lock leak in ipfw_commit_rules().
Tue, Jan 6, 5:51 PM
glebius accepted D54561: in6: Add a helper function to compute expiry times.
Tue, Jan 6, 4:29 PM

Mon, Jan 5

glebius requested changes to D54537: netinet: Use ICMP_REDIRECT_DEFAULT_TIMEOUT for redirtimeout and clarify sysctl description.

But where is the ICMP_REDIRECT_DEFAULT_TIMEOUT defined? The patch as is can't be compilable.

Mon, Jan 5, 8:22 PM
glebius accepted D54534: tcp: Unifidef use of rss software hash in syncache.

Thanks!

Mon, Jan 5, 7:47 PM
glebius requested review of D54535: ipfw: make the upper half lock sleepable.
Mon, Jan 5, 7:39 PM
glebius committed rG161db3a910a4: ipfw: make create_objects_compat() static.
Mon, Jan 5, 6:05 PM
glebius committed rGbc9c45df1c29: ipfw: use sparse initializer for dynamic opcodes.
Mon, Jan 5, 6:05 PM
glebius committed rGe4e30d5a8136: ipfw: remove extraneous argument from ipfw_add_protected_rule().
Mon, Jan 5, 6:04 PM

Thu, Jan 1

glebius closed D54089: net: attach IPv4 and IPv6 stacks to an interface with EVENTHANDLER(9).
Thu, Jan 1, 3:37 PM

Sat, Dec 27

glebius added a comment to D42972: netlink: Don't directly access ifnet members.

Well I think you're right. if_getifaddr() ( wrapper around ifp->if_addr ) is the right accessor to get link level address. Using that directly is much simpler than a iterator.

Sat, Dec 27, 6:38 PM

Dec 23 2025

glebius added a comment to D54356: iflib: Drop tx lock when freeing mbufs using simple_transmit.

Have you tried chaining mbufs instead of having an array? Does that have negative cache effect?

Dec 23 2025, 10:30 PM
glebius committed rG1aea5b0ba76a: nvme: make the interface standard file.
Dec 23 2025, 5:53 PM

Dec 22 2025

glebius committed rG88b04633c29e: libpcap: fix resource leaks and set error messages (authored by guy_alum.mit.edu).
Dec 22 2025, 2:24 AM
glebius committed rG129e15d49943: ipfw: fix bpf tap point lookup.
Dec 22 2025, 2:24 AM
glebius committed rGb275907fa338: if_tuntap: use ifnet_rename_event instead of ifnet_arrival_event.
Dec 22 2025, 2:24 AM
glebius committed rG0bd0c3295ac0: ng_ether: refactor to use interface EVENTHANDLER(9)s.
Dec 22 2025, 2:24 AM
glebius committed rG349fcf079ca3: net: add ifnet_rename_event EVENTHANDLER(9) for interface renaming.
Dec 22 2025, 2:24 AM
glebius committed rGe3d6cf8c3f6c: pf: make eventhandler_tag's static.
Dec 22 2025, 2:24 AM
glebius committed rG77939d64f23d: net: on interface detach purge multicast addresses after protocols.
Dec 22 2025, 2:24 AM

Dec 19 2025

glebius accepted D54306: netinet6: Disallow connections to IN6ADDR_ANY.

I'd also suggest to change it to SYSCTL_BOOL.

Dec 19 2025, 7:48 PM
glebius committed rGfe53a8a82c81: bpf: add missing IFT_BRIDGE in the write method.
Dec 19 2025, 7:24 PM

Dec 18 2025

glebius committed rG0d469d23715d: net: attach IPv4 and IPv6 stacks to an interface with EVENTHANDLER(9).
Dec 18 2025, 7:18 PM
glebius closed D54085: net: split ifnet_arrival_event into two events.
Dec 18 2025, 7:17 PM
glebius committed rG713b57c414b7: net: split ifnet_arrival_event into two events.
Dec 18 2025, 7:17 PM
glebius committed rGa4d766caf711: bpf: add a crutch to support if_vmove.
Dec 18 2025, 4:47 PM
glebius added inline comments to D54089: net: attach IPv4 and IPv6 stacks to an interface with EVENTHANDLER(9).
Dec 18 2025, 12:26 AM

Dec 17 2025

glebius committed rG104827151e2a: ng_iface: don't recursively enter epoch in the rcvdata method.
Dec 17 2025, 10:06 PM
glebius closed D54241: ng_iface: don't recursively enter epoch in the rcvdata method.
Dec 17 2025, 10:06 PM
glebius committed rGd4062b9f16e4: vlan: plug a new panic associated with interface removal.
Dec 17 2025, 9:16 PM
glebius committed rG0bf42a0a05b9: bpf: virtualize bpf_iflist.
Dec 17 2025, 8:47 PM
glebius accepted D54272: sysutils/cpu-microcode-amd: Update to 2025-12-02 snapshot.

As we discussed in private email exchange, this is speculatively supposed to be safe. I tested only on CPU Family=0x19 Model=0xa0 Stepping=0x02, that doesn't have updated BIOS. The microcode in the CPU matches the first version in the blob and safely failed to upgrade to the newer version. The expected failure message was printed.

Dec 17 2025, 6:59 PM

Dec 16 2025

glebius updated the diff for D53871: bpf: don't recursively enter the network epoch in bpf taps.
  • Rebase. USB already has been taken care of.
Dec 16 2025, 9:33 PM
glebius updated the diff for D54089: net: attach IPv4 and IPv6 stacks to an interface with EVENTHANDLER(9).
  • Rebase better.
Dec 16 2025, 8:51 PM
glebius updated the diff for D54085: net: split ifnet_arrival_event into two events.
  • Add missing line.
Dec 16 2025, 8:50 PM
glebius updated the diff for D54089: net: attach IPv4 and IPv6 stacks to an interface with EVENTHANDLER(9).
  • Rebased. No functional changes.
Dec 16 2025, 7:37 PM
glebius updated the diff for D54085: net: split ifnet_arrival_event into two events.
  • Split into two events
Dec 16 2025, 7:37 PM
glebius retitled D54085: net: split ifnet_arrival_event into two events from net: invoke ifnet_arrival_event handlers before linking in
Dec 16 2025, 7:36 PM
glebius committed rG8b780f75f7ef: libpcap: disable more code related to USB pseudo-interfaces.
Dec 16 2025, 6:26 PM