Jakub Kicinski says:
====================
tools: ynl: stop using libmnl
There is no strong reason to stop using libmnl in ynl but there
are a few small ones which add up.
First (as I remembered immediately after hitting send on v1),
C++ compilers do not like the libmnl for_each macros.
I haven't tried it myself, but having all the code directly
in YNL makes it easier for folks porting to C++ to modify them
and/or make YNL more C++ friendly.
Second, we do much more advanced netlink level parsing in ynl
than libmnl so it's hard to say that libmnl abstracts much from us.
The fact that this series, removing the libmnl dependency, only
adds <300 LoC shows that code savings aren't huge.
OTOH when new types are added (e.g. auto-int) we need to add
compatibility to deal with older version of libmnl (in fact,
even tho patches have been sent months ago, auto-ints are still
not supported in libmnl.git).
Thrid, the dependency makes ynl less self contained, and harder
to vendor in. Whether vendoring libraries into projects is a good
idea is a separate discussion, nonetheless, people want to do it.
Fourth, there are small annoyances with the libmnl APIs which
are hard to fix in backward-compatible ways. See the last patch
for example.
All in all, libmnl is a great library, but with all the code
generation and structured parsing, ynl is better served by going
its own way.
v1: https://lore.kernel.org/all/20240222235614.180876-1-kuba@kernel.org/
====================
Link: https://lore.kernel.org/r/20240227223032.1835527-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
To stick to libmnl wrappers in the past we had to use poll()
to check if there are any outstanding notifications on the socket.
This is no longer necessary, we can use MSG_DONTWAIT.
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/20240227223032.1835527-16-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Most libmnl socket helpers can be replaced by direct calls to
the underlying libc API. We need portid, the netlink manpage
suggests we bind() address of zero.
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/20240227223032.1835527-14-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
All YNL parsing callbacks take struct ynl_parse_arg as the argument.
Make that official by using a local callback type instead of mnl_cb_t.
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/20240227223032.1835527-12-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
There's only one set of callbacks in YNL, for netlink control
messages, and most of them are trivial. So implement the message
walking directly without depending on mnl_cb_run2().
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/20240227223032.1835527-11-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
ynl_recv_ack() is simple and it's the only user of mnl_cb_run().
Now that ynl_sock_read_msgs() exists it's actually less code
to use ynl_sock_read_msgs() instead of being special.
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/20240227223032.1835527-10-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
All callers to mnl_cb_run2() call mnl_socket_recvfrom() right before.
Wrap the two in a helper, take typed arguments (struct ynl_parse_arg),
instead of hoping that all callers remember that parser error handling
requires yarg.
In case of ynl_sock_read_family() we will no longer check for kernel
returning no data, but that would be a kernel bug, not worth complicating
the code to catch this. Calling mnl_cb_run2() on an empty buffer
is legal and results in STOP (1).
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/20240227223032.1835527-9-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Commit f2ba1e5e22 ("tools: ynl-gen: stop generating common notification handlers")
removed the last caller of the parse_cb_run() helper.
We no longer need to export ynl_cb_array.
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/20240227223032.1835527-8-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
All YNL parsing code expects a pointer to struct ynl_parse_arg AKA yarg.
For dump was pass in struct ynl_dump_state, which works fine, because
struct ynl_dump_state and struct ynl_parse_arg have identical layout
for the members that matter.. but it's a bit hacky.
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/20240227223032.1835527-7-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Create helpers for accessing payloads of struct nlmsg.
Use them instead of the libmnl ones.
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/20240227223032.1835527-5-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Don't use mnl attr helpers, we're trying to remove the libmnl
dependency. Create both signed and unsigned helpers, libmnl
had unsigned helpers, so code generator no longer needs
the mnl_type() hack.
The new helpers are written from first principles, but are
hopefully not too buggy.
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/20240227223032.1835527-3-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The temporary auto-int helpers are not really correct.
We can't treat signed and unsigned ints the same when
determining whether we need full 8B. I realized this
before sending the patch to add support in libmnl.
Unfortunately, that patch has not been merged,
so time to fix our local helpers. Use the mnl* name
for now, subsequent patches will address that.
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Reviewed-by: Donald Hunter <donald.hunter@gmail.com>
Link: https://lore.kernel.org/r/20240227223032.1835527-2-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
We never increment the group number iterator, so all groups
get recorded into index 0 of the mcast_groups[] array.
As a result YNL can only handle using the last group.
For example using the "netdev" sample on kernel with
page pool commands results in:
$ ./samples/netdev
YNL: Multicast group 'mgmt' not found
Most families have only one multicast group, so this hasn't
been noticed. Plus perhaps developers usually test the last
group which would have worked.
Fixes: 86878f14d7 ("tools: ynl: user space helpers")
Reviewed-by: Donald Hunter <donald.hunter@gmail.com>
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/20240226214019.1255242-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Since commit 7c59c9c8f2 ("tools: ynl: generate code for ovs families")
we need relatively recent OvS headers to get YNL to compile.
Add the direct include workaround to fix compilation on less
up-to-date OSes like CentOS 9.
Reviewed-by: Donald Hunter <donald.hunter@gmail.com>
Link: https://lore.kernel.org/r/20240226225806.1301152-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add test case for multicast packet confirm race.
Without preceding patch, this should result in:
WARNING: CPU: 0 PID: 38 at net/netfilter/nf_conntrack_core.c:1198 __nf_conntrack_confirm+0x3ed/0x5f0
Workqueue: events_unbound macvlan_process_broadcast
RIP: 0010:__nf_conntrack_confirm+0x3ed/0x5f0
? __nf_conntrack_confirm+0x3ed/0x5f0
nf_confirm+0x2ad/0x2d0
nf_hook_slow+0x36/0xd0
ip_local_deliver+0xce/0x110
__netif_receive_skb_one_core+0x4f/0x70
process_backlog+0x8c/0x130
[..]
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
conntrack nf_confirm logic cannot handle cloned skbs referencing
the same nf_conn entry, which will happen for multicast (broadcast)
frames on bridges.
Example:
macvlan0
|
br0
/ \
ethX ethY
ethX (or Y) receives a L2 multicast or broadcast packet containing
an IP packet, flow is not yet in conntrack table.
1. skb passes through bridge and fake-ip (br_netfilter)Prerouting.
-> skb->_nfct now references a unconfirmed entry
2. skb is broad/mcast packet. bridge now passes clones out on each bridge
interface.
3. skb gets passed up the stack.
4. In macvlan case, macvlan driver retains clone(s) of the mcast skb
and schedules a work queue to send them out on the lower devices.
The clone skb->_nfct is not a copy, it is the same entry as the
original skb. The macvlan rx handler then returns RX_HANDLER_PASS.
5. Normal conntrack hooks (in NF_INET_LOCAL_IN) confirm the orig skb.
The Macvlan broadcast worker and normal confirm path will race.
This race will not happen if step 2 already confirmed a clone. In that
case later steps perform skb_clone() with skb->_nfct already confirmed (in
hash table). This works fine.
But such confirmation won't happen when eb/ip/nftables rules dropped the
packets before they reached the nf_confirm step in postrouting.
Pablo points out that nf_conntrack_bridge doesn't allow use of stateful
nat, so we can safely discard the nf_conn entry and let inet call
conntrack again.
This doesn't work for bridge netfilter: skb could have a nat
transformation. Also bridge nf prevents re-invocation of inet prerouting
via 'sabotage_in' hook.
Work around this problem by explicit confirmation of the entry at LOCAL_IN
time, before upper layer has a chance to clone the unconfirmed entry.
The downside is that this disables NAT and conntrack helpers.
Alternative fix would be to add locking to all code parts that deal with
unconfirmed packets, but even if that could be done in a sane way this
opens up other problems, for example:
-m physdev --physdev-out eth0 -j SNAT --snat-to 1.2.3.4
-m physdev --physdev-out eth1 -j SNAT --snat-to 1.2.3.5
For multicast case, only one of such conflicting mappings will be
created, conntrack only handles 1:1 NAT mappings.
Users should set create a setup that explicitly marks such traffic
NOTRACK (conntrack bypass) to avoid this, but we cannot auto-bypass
them, ruleset might have accept rules for untracked traffic already,
so user-visible behaviour would change.
Suggested-by: Pablo Neira Ayuso <pablo@netfilter.org>
Fixes: 1da177e4c3 ("Linux-2.6.12-rc2")
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=217777
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Use function names that match the implementation in kernel-doc comments
to avoid kernel-doc warnings:
dpll3xxx.c:938: warning: expecting prototype for omap3_non_core_dpll_save_context(). Prototype was for omap3_noncore_dpll_save_context() instead
dpll3xxx.c:967: warning: expecting prototype for omap3_core_dpll_restore_context(). Prototype was for omap3_noncore_dpll_restore_context() instead
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Cc: Tero Kristo <kristo@kernel.org>
Cc: linux-omap@vger.kernel.org
Cc: Michael Turquette <mturquette@baylibre.com>
Cc: Stephen Boyd <sboyd@kernel.org>
Cc: linux-clk@vger.kernel.org
Link: https://lore.kernel.org/r/20240115054739.4988-1-rdunlap@infradead.org
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Commit d0009effa8 ("netfilter: nf_tables: validate NFPROTO_* family") added
some validation of NFPROTO_* families in the nft_compat module, but it broke
the ability to use legacy iptables modules in dual-stack nftables.
While with legacy iptables one had to independently manage IPv4 and IPv6
tables, with nftables it is possible to have dual-stack tables sharing the
rules. Moreover, it was possible to use rules based on legacy iptables
match/target modules in dual-stack nftables.
As an example, the program from [2] creates an INET dual-stack family table
using an xt_bpf based rule, which looks like the following (the actual output
was generated with a patched nft tool as the current nft tool does not parse
dual stack tables with legacy match rules, so consider it for illustrative
purposes only):
table inet testfw {
chain input {
type filter hook prerouting priority filter; policy accept;
bytecode counter packets 0 bytes 0 accept
}
}
After d0009effa8 ("netfilter: nf_tables: validate NFPROTO_* family") we get
EOPNOTSUPP for the above program.
Fix this by allowing NFPROTO_INET for nft_(match/target)_validate(), but also
restrict the functions to classic iptables hooks.
Changes in v3:
* clarify that upstream nft will not display such configuration properly and
that the output was generated with a patched nft tool
* remove example program from commit description and link to it instead
* no code changes otherwise
Changes in v2:
* restrict nft_(match/target)_validate() to classic iptables hooks
* rewrite example program to use unmodified libnftnl
Fixes: d0009effa8 ("netfilter: nf_tables: validate NFPROTO_* family")
Link: https://lore.kernel.org/all/Zc1PfoWN38UuFJRI@calendula/T/#mc947262582c90fec044c7a3398cc92fac7afea72 [1]
Link: https://lore.kernel.org/all/20240220145509.53357-1-ignat@cloudflare.com/ [2]
Reported-by: Jordan Griege <jgriege@cloudflare.com>
Signed-off-by: Ignat Korchagin <ignat@cloudflare.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
The pointer div_addr is being assigned a value that is never used, it is
being re-assigned a different value near the end of the function where
it is being read in the next statement. The initialization is redundant
and can be removed.
Cleans up clang scan build warning:
drivers/clk/xilinx/clk-xlnx-clock-wizard.c:501:16: warning: Value stored
to 'div_addr' during its initialization is never read [deadcode.DeadStores]
Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Link: https://lore.kernel.org/r/20240223134347.3908301-1-colin.i.king@gmail.com
Reviewed-by: Michal Simek <michal.simek@amd.com>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Commit eaae75754d ("docs: turn off "smart quotes" in the HTML build")
disabled conversion of quote marks along with that of dashes.
Despite the short summary, the change affects not only HTML build
but also other build targets including PDF.
However, as "smart quotes" had been enabled for more than half a
decade already, quite a few readers of HTML pages are likely expecting
conversions of "foo" -> “foo” and 'bar' -> ‘bar’.
Furthermore, in LaTeX typesetting convention, it is common to use
distinct marks for opening and closing quote marks.
To satisfy such readers' expectation, restore conversion of quotes
only by setting smartquotes_action [1].
Link: [1] https://www.sphinx-doc.org/en/master/usage/configuration.html#confval-smartquotes_action
Cc: stable@vger.kernel.org # v6.4
Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Link: https://lore.kernel.org/r/20240225094600.65628-1-akiyks@gmail.com
[WHY]
Some eDP panels' ext caps don't write initial values. The value of
dpcd_addr (0x317) can be random and the backlight control interface
will be incorrect.
[HOW]
Add new panel patches to remove sink ext caps.
Cc: Mario Limonciello <mario.limonciello@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.5.x
Cc: Tsung-hua Lin <tsung-hua.lin@amd.com>
Cc: Chris Chi <moukong.chi@amd.com>
Reviewed-by: Wayne Lin <wayne.lin@amd.com>
Acked-by: Alex Hung <alex.hung@amd.com>
Signed-off-by: Ryan Lin <tsung-hua.lin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
clkdev DEV ID information is limited to an array of 20 bytes
(MAX_DEV_ID). It is possible that the ID could be longer than
that. If so, the lookup will fail because the "real ID" will
not match the copied value.
For instance, generating a device name for the I2C Designware
module using the PCI ID can result in a name of:
i2c_designware.39424
clkdev_create() will store:
i2c_designware.3942
The stored name is one off and will not match correctly during probe.
Increase the size of the ID to allow for a longer name.
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Link: https://lore.kernel.org/r/20240223202556.2194021-1-michael.j.ruhl@intel.com
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
The adev can be found from bo by amdgpu_ttm_adev(bo->tbo.bdev),
and adev is also not used in the function
amdgpu_amdkfd_map_gtt_bo_to_gart().
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Fix mask used for esm ctrl register to get pcie link
speed on smu_v11_0_3, smu_v13_0_2 & smu_v13_0_6
Fixes: 511a95552e ("drm/amd/pm: Add SMU 13.0.6 support")
Fixes: c05d1c4015 ("drm/amd/swsmu: add aldebaran smu13 ip support (v3)")
Fixes: f1c3785931 ("drm/amd/powerplay: add Arcturus support for gpu metrics export")
Signed-off-by: Asad Kamal <asad.kamal@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Skip reporting pcie link width/speed on vfs for
smu_v13_0_6 & smu_v13_0_2
Signed-off-by: Asad Kamal <asad.kamal@amd.com>
Reviewed-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
v1:
enabel pp_od_clk_voltage node for gfx 9.4.3 SRIOV and BM.
v2:
add onevf check for gfx 9.4.3
v3:
refine code check order to make function clearly.
Signed-off-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
When building with CONFIG_SYSFS=n, the build error below is triggered:
ld: drivers/power/supply/power_supply_core.o:(.data+0x0): undefined
reference to `power_supply_attr_group'
The problem is that power_supply_attr_group is needed in
power_supply_core.c but defined in power_supply_sysfs.c, which is only
targeted with CONFIG_SYSFS=y. Therefore, move the extern declaration into
the #ifdef block that checks for CONFIG_SYSFS, and define an empty static
const struct otherwise. This is safe because the macro __ATRIBUTE_GROUPS in
power_supply_core.c will expand into an empty attribute_group array.
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reported-by: Nathan Chancellor <nathan@kernel.org>
Link: https://lore.kernel.org/all/20240227214916.GA3699076@dev-arch.thelio-3990X/
Fixes: 7b46b60944 ("power: supply: core: constify the struct device_type usage")
Signed-off-by: Ricardo B. Marliere <ricardo@marliere.net>
Tested-by: Nathan Chancellor <nathan@kernel.org> # build
Link: https://lore.kernel.org/r/20240228-device_cleanup-power-v1-1-52c0321c48e1@marliere.net
Signed-off-by: Sebastian Reichel <sebastian.reichel@collabora.com>
rk3588 syscon clock fixes and removal of overall clock-number
from the rk3588 binding header and a prerequisite for later
improvements to the rk3588 linked clocks.
-----BEGIN PGP SIGNATURE-----
iQFEBAABCAAuFiEE7v+35S2Q1vLNA3Lx86Z5yZzRHYEFAmXe7ScQHGhlaWtvQHNu
dGVjaC5kZQAKCRDzpnnJnNEdgdLPCACLFH4I+aJ8OV/SuKW+NroDpgss39x082YV
wpYp3Q+JEh7K2ZoESmvaqm1Fi5B5bCSo10LBgcC343S/rkmQGrXKSRYScfMDVR6K
GGSz/o+Aub4WdUFjsZbj+lI97XslFuReBSNQftz5ZgNXH1HObX69dqAJXhX6Pdmg
nbQ92PG8FJxjOlRvJGOhhs91c7UQwPWEY87xsJze2Vbv3oYZzbDVuXRtjwB20OMP
J0fIpRSZvLd/89s6BamOKyKXkBuAJuuoF9S3jm54/EDcqkZO6hY7fTR95jSTOmHh
eE3vc2F5G3kHWjNFIvqlA7izzlavd150qFYwqi6LGstwbdQWg/Lp
=SPXE
-----END PGP SIGNATURE-----
Merge tag 'v6.9-rockchip-clk1' of git://git.kernel.org/pub/scm/linux/kernel/git/mmind/linux-rockchip into clk-rockchip
Pull Rockchip clk driver updates from Heiko Stuebner:
- New pll-rate for rk3568
- i2s rate improvements for rk3399
- rk3588 syscon clock fixes and removal of overall clock-number from
the rk3588 binding header
- a prerequisite for later improvements to the rk3588 linked clocks
* tag 'v6.9-rockchip-clk1' of git://git.kernel.org/pub/scm/linux/kernel/git/mmind/linux-rockchip:
clk: rockchip: rk3399: Allow to set rate of clk_i2s0_frac's parent
clk: rockchip: rk3588: use linked clock ID for GATE_LINK
clk: rockchip: rk3588: fix indent
clk: rockchip: rk3588: fix pclk_vo0grf and pclk_vo1grf
dt-bindings: clock: rk3588: add missing PCLK_VO1GRF
dt-bindings: clock: rk3588: drop CLK_NR_CLKS
clk: rockchip: rk3588: fix CLK_NR_CLKS usage
clk: rockchip: rk3568: Add PLL rate for 128MHz
When building without CONFIG_SYSFS, there is an error because of a
recent refactoring that failed to update the stub of
power_supply_init_attrs():
drivers/power/supply/power_supply_core.c: In function 'power_supply_class_init':
drivers/power/supply/power_supply_core.c:1630:9: error: too few arguments to function 'power_supply_init_attrs'
1630 | power_supply_init_attrs();
| ^~~~~~~~~~~~~~~~~~~~~~~
In file included from drivers/power/supply/power_supply_core.c:25:
drivers/power/supply/power_supply.h:25:20: note: declared here
25 | static inline void power_supply_init_attrs(struct device_type *dev_type) {}
| ^~~~~~~~~~~~~~~~~~~~~~~
Update the stub function to take no parameters like the rest of the
refactoring, which resolves the build error.
Fixes: 7b46b60944 ("power: supply: core: constify the struct device_type usage")
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Reviewed-by: Ricardo B. Marliere <ricardo@marliere.net>
Link: https://lore.kernel.org/r/20240227-fix-power_supply_init_attrs-stub-v1-1-43365e68d4b3@kernel.org
Signed-off-by: Sebastian Reichel <sebastian.reichel@collabora.com>
Puranjay Mohan says:
====================
bpf, arm64: use BPF prog pack allocator in BPF JIT
Changes in V8 => V9:
V8: https://lore.kernel.org/bpf/20240221145106.105995-1-puranjay12@gmail.com/
1. Rebased on bpf-next/master
2. Added Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Changes in V7 => V8:
V7: https://lore.kernel.org/bpf/20240125133159.85086-1-puranjay12@gmail.com/
1. Rebase on bpf-next/master
2. Fix __text_poke() by removing usage of 'ret' that was never set.
Changes in V6 => V7:
V6: https://lore.kernel.org/all/20240124164917.119997-1-puranjay12@gmail.com/
1. Rebase on bpf-next/master.
Changes in V5 => V6:
V5: https://lore.kernel.org/all/20230908144320.2474-1-puranjay12@gmail.com/
1. Implement a text poke api to reduce code repeatition.
2. Use flush_icache_range() in place of caches_clean_inval_pou() in the
functions that modify code.
3. Optimize the bpf_jit_free() by not copying the all instructions on
the rw image to the ro_image
Changes in V4 => v5:
1. Remove the patch for making prog pack allocator portable as it will come
through the RISCV tree[1].
2. Add a new function aarch64_insn_set() to be used in
bpf_arch_text_invalidate() for putting illegal instructions after a
program is removed. The earlier implementation of bpf_arch_text_invalidate()
was calling aarch64_insn_patch_text_nosync() in a loop and making it slow
because each call invalidated the cache.
Here is test_tag now:
[root@ip-172-31-6-176 bpf]# time ./test_tag
test_tag: OK (40945 tests)
real 0m19.695s
user 0m1.514s
sys 0m17.841s
test_tag without these patches:
[root@ip-172-31-6-176 bpf]# time ./test_tag
test_tag: OK (40945 tests)
real 0m21.487s
user 0m1.647s
sys 0m19.106s
test_tag in the previous version was really slow > 2 minutes. see [2]
3. Add cache invalidation in aarch64_insn_copy() so other users can call the
function without worrying about the cache. Currently only bpf_arch_text_copy()
is using it, but there might be more users in the future.
Chanes in V3 => V4: Changes only in 3rd patch
1. Fix the I-cache maintenance: Clean the data cache and invalidate the i-Cache
only *after* the instructions have been copied to the ROX region.
Chanes in V2 => V3: Changes only in 3rd patch
1. Set prog = orig_prog; in the failure path of bpf_jit_binary_pack_finalize()
call.
2. Add comments explaining the usage of the offsets in the exception table.
Changes in v1 => v2:
1. Make the naming consistent in the 3rd patch:
ro_image and image
ro_header and header
ro_image_ptr and image_ptr
2. Use names dst/src in place of addr/opcode in second patch.
3. Add Acked-by: Song Liu <song@kernel.org> in 1st and 2nd patch.
BPF programs currently consume a page each on ARM64. For systems with many BPF
programs, this adds significant pressure to instruction TLB. High iTLB pressure
usually causes slow down for the whole system.
Song Liu introduced the BPF prog pack allocator[3] to mitigate the above issue.
It packs multiple BPF programs into a single huge page. It is currently only
enabled for the x86_64 BPF JIT.
This patch series enables the BPF prog pack allocator for the ARM64 BPF JIT.
====================================================
Performance Analysis of prog pack allocator on ARM64
====================================================
To test the performance of the BPF prog pack allocator on ARM64, a stresser
tool[4] was built. This tool loads 8 BPF programs on the system and triggers
5 of them in an infinite loop by doing system calls.
The runner script starts 20 instances of the above which loads 8*20=160 BPF
programs on the system, 5*20=100 of which are being constantly triggered.
In the above environment we try to build Python-3.8.4 and try to find different
iTLB metrics for the compilation done by gcc-12.2.0.
The source code[5] is configured with the following command:
./configure --enable-optimizations --with-ensurepip=install
Then the runner script is executed with the following command:
./run.sh "perf stat -e ITLB_WALK,L1I_TLB,INST_RETIRED,iTLB-load-misses -a make -j32"
This builds Python while 160 BPF programs are loaded and 100 are being constantly
triggered and measures iTLB related metrics.
The output of the above command is discussed below before and after enabling the
BPF prog pack allocator.
The tests were run on qemu-system-aarch64 with 32 cpus, 4G memory, -machine virt,
-cpu host, and -enable-kvm.
Results
-------
Before enabling prog pack allocator:
------------------------------------
Performance counter stats for 'system wide':
333278635 ITLB_WALK
6762692976558 L1I_TLB
25359571423901 INST_RETIRED
15824054789 iTLB-load-misses
189.029769053 seconds time elapsed
After enabling prog pack allocator:
-----------------------------------
Performance counter stats for 'system wide':
190333544 ITLB_WALK
6712712386528 L1I_TLB
25278233304411 INST_RETIRED
5716757866 iTLB-load-misses
185.392650561 seconds time elapsed
Improvements in metrics
-----------------------
Compilation time ---> 1.92% faster
iTLB-load-misses/Sec (Less is better) ---> 63.16% decrease
ITLB_WALK/1000 INST_RETIRED (Less is better) ---> 42.71% decrease
ITLB_Walk/L1I_TLB (Less is better) ---> 42.47% decrease
[1] https://git.kernel.org/pub/scm/linux/kernel/git/riscv/linux.git/commit/?h=for-next&id=20e490adea279d49d57b800475938f5b67926d98
[2] https://lore.kernel.org/all/CANk7y0gcP3dF2mESLp5JN1+9iDfgtiWRFGqLkCgZD6wby1kQOw@mail.gmail.com/
[3] https://lore.kernel.org/bpf/20220204185742.271030-1-song@kernel.org/
[4] https://github.com/puranjaymohan/BPF-Allocator-Bench
[5] https://www.python.org/ftp/python/3.8.4/Python-3.8.4.tgz
====================
Link: https://lore.kernel.org/r/20240228141824.119877-1-puranjay12@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Use bpf_jit_binary_pack_alloc for memory management of JIT binaries in
ARM64 BPF JIT. The bpf_jit_binary_pack_alloc creates a pair of RW and RX
buffers. The JIT writes the program into the RW buffer. When the JIT is
done, the program is copied to the final RX buffer
with bpf_jit_binary_pack_finalize.
Implement bpf_arch_text_copy() and bpf_arch_text_invalidate() for ARM64
JIT as these functions are required by bpf_jit_binary_pack allocator.
Signed-off-by: Puranjay Mohan <puranjay12@gmail.com>
Acked-by: Song Liu <song@kernel.org>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Link: https://lore.kernel.org/r/20240228141824.119877-3-puranjay12@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
The text_poke API is used to implement functions like memcpy() and
memset() for instruction memory (RO+X). The implementation is similar to
the x86 version.
This will be used by the BPF JIT to write and modify BPF programs. There
could be more users of this in the future.
Signed-off-by: Puranjay Mohan <puranjay12@gmail.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Link: https://lore.kernel.org/r/20240228141824.119877-2-puranjay12@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Setting the uniquifier for fscache via the fsc= mount
option is currently broken in NFSv4.
Fix this by passing fscache_uniq to root_fc if possible.
Signed-off-by: Chen Hanxiao <chenhx.fnst@fujitsu.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
While decoding filelayout getdeviceinfo received, print out the
information about the location of data servers (IPs).
Generic getdeviceinfo tracepoints prints the MDS's ip for the
dstaddr. In this patch, separate the MDS's address from the
DS's addresses.
Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Switch order of operations to avoid creating a short XDR buffer:
e.g., buflen = 12, old xdrlen = 12, new xdrlen = 20.
Having a short XDR buffer leads to lxa_maxcount be a few bytes
less than what is needed to retrieve the whole list when using
a buflen as returned by a call with size = 0:
buflen = listxattr(path, NULL, 0);
buf = malloc(buflen);
buflen = listxattr(path, buf, buflen);
For a file with one attribute (name = '123456'), the first call
with size = 0 will return buflen = 12 ('user.123456\x00').
The second call with size = 12, sends LISTXATTRS with
lxa_maxcount = 12 + 8 (cookie) + 4 (array count) = 24. The
XDR buffer needs 8 (cookie) + 4 (array count) + 4 (name count)
+ 6 (name len) + 2 (padding) + 4 (eof) = 28 which is 4 bytes
shorter than the lxa_maxcount provided in the call.
Fixes: 04a5da690e ("NFSv4.2: define limits and sizes for user xattr handling")
Signed-off-by: Jorge Mora <mora@netapp.com>
Reviewed-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
A call to listxattr() with a buffer size = 0 returns the actual
size of the buffer needed for a subsequent call. When size > 0,
nfs4_listxattr() does not return an error because either
generic_listxattr() or nfs4_listxattr_nfs4_label() consumes
exactly all the bytes then size is 0 when calling
nfs4_listxattr_nfs4_user() which then triggers the following
kernel BUG:
[ 99.403778] kernel BUG at mm/usercopy.c:102!
[ 99.404063] Internal error: Oops - BUG: 00000000f2000800 [#1] SMP
[ 99.408463] CPU: 0 PID: 3310 Comm: python3 Not tainted 6.6.0-61.fc40.aarch64 #1
[ 99.415827] Call trace:
[ 99.415985] usercopy_abort+0x70/0xa0
[ 99.416227] __check_heap_object+0x134/0x158
[ 99.416505] check_heap_object+0x150/0x188
[ 99.416696] __check_object_size.part.0+0x78/0x168
[ 99.416886] __check_object_size+0x28/0x40
[ 99.417078] listxattr+0x8c/0x120
[ 99.417252] path_listxattr+0x78/0xe0
[ 99.417476] __arm64_sys_listxattr+0x28/0x40
[ 99.417723] invoke_syscall+0x78/0x100
[ 99.417929] el0_svc_common.constprop.0+0x48/0xf0
[ 99.418186] do_el0_svc+0x24/0x38
[ 99.418376] el0_svc+0x3c/0x110
[ 99.418554] el0t_64_sync_handler+0x120/0x130
[ 99.418788] el0t_64_sync+0x194/0x198
[ 99.418994] Code: aa0003e3 d000a3e0 91310000 97f49bdb (d4210000)
Issue is reproduced when generic_listxattr() returns 'system.nfs4_acl',
thus calling lisxattr() with size = 16 will trigger the bug.
Add check on nfs4_listxattr() to return ERANGE error when it is
called with size > 0 and the return value is greater than size.
Fixes: 012a211abd ("NFSv4.2: hook in the user extended attribute handlers")
Signed-off-by: Jorge Mora <mora@netapp.com>
Reviewed-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
With this patch, mount command will show fsc=xxx if set:
If -o fsc=6666
clientaddr=192.168.122.208,fsc=6666,local_lock=none
If only -o fsc
clientaddr=192.168.122.208,fsc,local_lock=none
Signed-off-by: Chen Hanxiao <chenhx.fnst@fujitsu.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
In order to get the latency per xprt under the same clientid this patch
adds xprt_id to the tracepoint output.
Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
Tested-by: Chen Hanxiao <chenhx.fnst@fujitsu.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
The intent is to check if the strings' are truncated or not. So, >= should
be used instead of >, because strlcat() and snprintf() return the length of
the output, excluding the trailing NULL.
Fixes: a02d692611 ("SUNRPC: Provide functions for managing universal addresses")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Allocate the common ucall pool using vm_vaddr_alloc_shared() so that the
ucall structures will be placed in shared (unencrypted) memory for VMs
with support for protected (encrypted) memory, e.g. x86's SEV.
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Sean Christopherson <seanjc@google.com>
Cc: Vishal Annapurve <vannapurve@google.com>
Cc: Ackerly Tng <ackerleytng@google.com>
cc: Andrew Jones <andrew.jones@linux.dev>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: Michael Roth <michael.roth@amd.com>
Tested-by: Carlos Bilbao <carlos.bilbao@amd.com>
Signed-off-by: Peter Gonda <pgonda@google.com>
[sean: massage changelog]
Link: https://lore.kernel.org/r/20240223004258.3104051-7-seanjc@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
Test programs may wish to allocate shared vaddrs for things like
sharing memory with the guest. Since protected vms will have their
memory encrypted by default an interface is needed to explicitly
request shared pages.
Implement this by splitting the common code out from vm_vaddr_alloc()
and introducing a new vm_vaddr_alloc_shared().
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Sean Christopherson <seanjc@google.com>
Cc: Vishal Annapurve <vannapurve@google.com>
Cc: Ackerly Tng <ackerleytng@google.com>
cc: Andrew Jones <andrew.jones@linux.dev>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: Michael Roth <michael.roth@amd.com>
Reviewed-by: Itaru Kitayama <itaru.kitayama@fujitsu.com>
Tested-by: Carlos Bilbao <carlos.bilbao@amd.com>
Signed-off-by: Michael Roth <michael.roth@amd.com>
Signed-off-by: Peter Gonda <pgonda@google.com>
Link: https://lore.kernel.org/r/20240223004258.3104051-6-seanjc@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
Add support for differentiating between protected (a.k.a. private, a.k.a.
encrypted) memory and normal (a.k.a. shared) memory for VMs that support
protected guest memory, e.g. x86's SEV. Provide and manage a common
bitmap for tracking whether a given physical page resides in protected
memory, as support for protected memory isn't x86 specific, i.e. adding a
arch hook would be a net negative now, and in the future.
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Sean Christopherson <seanjc@google.com>
Cc: Vishal Annapurve <vannapurve@google.com>
Cc: Ackerley Tng <ackerleytng@google.com>
cc: Andrew Jones <andrew.jones@linux.dev>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: Michael Roth <michael.roth@amd.com>
Reviewed-by: Itaru Kitayama <itaru.kitayama@fujitsu.com>
Tested-by: Carlos Bilbao <carlos.bilbao@amd.com>
Originally-by: Michael Roth <michael.roth@amd.com>
Signed-off-by: Peter Gonda <pgonda@google.com>
Co-developed-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20240223004258.3104051-5-seanjc@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
Add sparsebit_for_each_set_range() to allow iterator over a range of set
bits in a range. This will be used by x86 SEV guests to process protected
physical pages (each such page needs to be encrypted _after_ being "added"
to the VM).
Tested-by: Carlos Bilbao <carlos.bilbao@amd.com>
Signed-off-by: Ackerley Tng <ackerleytng@google.com>
[sean: split to separate patch]
Link: https://lore.kernel.org/r/20240223004258.3104051-4-seanjc@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
Make all sparsebit struct pointers "const" where appropriate. This will
allow adding a bitmap to track protected/encrypted physical memory that
tests can access in a read-only fashion.
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Sean Christopherson <seanjc@google.com>
Cc: Vishal Annapurve <vannapurve@google.com>
Cc: Ackerley Tng <ackerleytng@google.com>
Cc: Andrew Jones <andrew.jones@linux.dev>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: Michael Roth <michael.roth@amd.com>
Tested-by: Carlos Bilbao <carlos.bilbao@amd.com>
Signed-off-by: Michael Roth <michael.roth@amd.com>
Signed-off-by: Peter Gonda <pgonda@google.com>
[sean: massage changelog]
Link: https://lore.kernel.org/r/20240223004258.3104051-3-seanjc@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
