mbox series

[v2,bpf,00/11] bpf: random unpopular userspace fixes (32 bit et al)

Message ID 20220421003152.339542-1-alobakin@pm.me (mailing list archive)
Headers show
Series bpf: random unpopular userspace fixes (32 bit et al) | expand

Message

Alexander Lobakin April 21, 2022, 12:38 a.m. UTC 
This mostly issues the cross build (1) errors for 32 bit (2)
MIPS (3) with minimal configuration (4) on Musl (5). The majority
of them aren't yesterday's, so it is a "who does need it outside
of x86_64 or ARM64?" moment again.
Trivial stuff in general, not counting the first three (they are
50/50).

From v1[0]:
 - use *___local struct definitions for BPF programs instead of
   BTF_TYPE_EMIT() and ifdef-play (Andrii);
 - cast uin64_t to unsigned long long to *really* fix the format
   literal warnings (Song, David, Andrii);
 - collect Acked-bys for the rest (Maciej, Kumar, Song);
 - adjust the subjects to match their usual look (Andrii);
 - expand the commit messages a bit for 0008
   (-Wshift-count-overflow) and 0010 (-Wsequence-point) a bit to
   mention they actually mitigate the third-party issues (Andrii);
 - rebase and send to bpf instead of bpf-next (hope the first three
   are okay for it).

[0] https://lore.kernel.org/bpf/20220414223704.341028-1-alobakin@pm.me

Alexander Lobakin (11):
  bpftool: use a local copy of perf_event to fix accessing ::bpf_cookie
  bpftool: define a local bpf_perf_link to fix accessing its fields
  bpftool: use a local bpf_perf_event_value to fix accessing its fields
  bpftool: fix fcntl.h include
  samples/bpf: add 'asm/mach-generic' include path for every MIPS
  samples/bpf: use host bpftool to generate vmlinux.h, not target
  samples/bpf: fix uin64_t format literals
  samples/bpf: fix false-positive right-shift underflow warnings
  samples/bpf: fix include order for non-Glibc environments
  samples/bpf: fix -Wsequence-point
  samples/bpf: xdpsock: fix -Wmaybe-uninitialized

 samples/bpf/Makefile                      |  7 +++---
 samples/bpf/cookie_uid_helper_example.c   | 12 +++++-----
 samples/bpf/lathist_kern.c                |  2 +-
 samples/bpf/lwt_len_hist_kern.c           |  2 +-
 samples/bpf/lwt_len_hist_user.c           |  7 +++---
 samples/bpf/task_fd_query_user.c          |  2 +-
 samples/bpf/test_lru_dist.c               |  3 ++-
 samples/bpf/tracex2_kern.c                |  2 +-
 samples/bpf/xdpsock_user.c                |  5 +++--
 tools/bpf/bpftool/skeleton/pid_iter.bpf.c | 15 ++++++++++---
 tools/bpf/bpftool/skeleton/profiler.bpf.c | 27 ++++++++++++++---------
 tools/bpf/bpftool/tracelog.c              |  2 +-
 12 files changed, 53 insertions(+), 33 deletions(-)

--
2.36.0

Comments

Alexei Starovoitov April 21, 2022, 12:40 a.m. UTC | #1 
On Wed, Apr 20, 2022 at 5:38 PM Alexander Lobakin <alobakin@pm.me> wrote:

Again?

-----BEGIN PGP MESSAGE-----
Version: ProtonMail

wcFMA165ASBBe6s8AQ/8C9y4TqXgASA5xBT7UIf2GyTQRjKWcy/6kT1dkjkF
FldAOhehhgLYjLJzNAIkecOQfz/XNapW3GdrQDq11pq9Bzs1SJJekGXlHVIW

Sorry I'm tossing the series out of patchwork.
Toke Høiland-Jørgensen April 21, 2022, 10:52 a.m. UTC | #2 
Alexei Starovoitov <alexei.starovoitov@gmail.com> writes:

> On Wed, Apr 20, 2022 at 5:38 PM Alexander Lobakin <alobakin@pm.me> wrote:
>
> Again?
>
> -----BEGIN PGP MESSAGE-----
> Version: ProtonMail
>
> wcFMA165ASBBe6s8AQ/8C9y4TqXgASA5xBT7UIf2GyTQRjKWcy/6kT1dkjkF
> FldAOhehhgLYjLJzNAIkecOQfz/XNapW3GdrQDq11pq9Bzs1SJJekGXlHVIW
>
> Sorry I'm tossing the series out of patchwork.

FWIW I'm not seeing this in the version I pulled from Lore. So maybe
it's something ProtonMail does on a per-recipient basis? Still really
weird to do behind the scenes, though... :/

-Toke
Alexander Lobakin April 21, 2022, 10:39 p.m. UTC | #3 
From: Alexei Starovoitov <alexei.starovoitov@gmail.com>
Date: Wed, 20 Apr 2022 17:40:34 -0700

> On Wed, Apr 20, 2022 at 5:38 PM Alexander Lobakin <alobakin@pm.me> wrote:
>
> Again?
>
> -----BEGIN PGP MESSAGE-----
> Version: ProtonMail
>
> wcFMA165ASBBe6s8AQ/8C9y4TqXgASA5xBT7UIf2GyTQRjKWcy/6kT1dkjkF
> FldAOhehhgLYjLJzNAIkecOQfz/XNapW3GdrQDq11pq9Bzs1SJJekGXlHVIW
>
> Sorry I'm tossing the series out of patchwork.

Oh sorry, I was hoping upgrading Bridge would help >_<

Let me know if you're reading this particular message in your inbox
finely. Toke guessed it precisely regarding the per-recipient lists
-- Proton by default saves every address I've ever sent mails to to
Contacts and then tries to fetch PGP public keys for each contact.
Again, for some reason, for a couple addresses, including
ast@kernel.org, it managed to fetch something, but that something
was sorta broken. So at the end I've been having broken PGP for
the address I've never manually set or even wanted PGP.
If it's still messed, I'll contact support then. Sorry again for
this.

Thanks,
Al
Toke Høiland-Jørgensen April 21, 2022, 11:17 p.m. UTC | #4 
Alexander Lobakin <alobakin@pm.me> writes:

> From: Alexei Starovoitov <alexei.starovoitov@gmail.com>
> Date: Wed, 20 Apr 2022 17:40:34 -0700
>
>> On Wed, Apr 20, 2022 at 5:38 PM Alexander Lobakin <alobakin@pm.me> wrote:
>>
>> Again?
>>
>> -----BEGIN PGP MESSAGE-----
>> Version: ProtonMail
>>
>> wcFMA165ASBBe6s8AQ/8C9y4TqXgASA5xBT7UIf2GyTQRjKWcy/6kT1dkjkF
>> FldAOhehhgLYjLJzNAIkecOQfz/XNapW3GdrQDq11pq9Bzs1SJJekGXlHVIW
>>
>> Sorry I'm tossing the series out of patchwork.
>
> Oh sorry, I was hoping upgrading Bridge would help >_<
>
> Let me know if you're reading this particular message in your inbox
> finely. Toke guessed it precisely regarding the per-recipient lists
> -- Proton by default saves every address I've ever sent mails to to
> Contacts and then tries to fetch PGP public keys for each contact.
> Again, for some reason, for a couple addresses, including
> ast@kernel.org, it managed to fetch something, but that something
> was sorta broken. So at the end I've been having broken PGP for
> the address I've never manually set or ev
> en wanted PGP.
> If it's still messed, I'll contact support then. Sorry again for
> this.

Heh, yeah, now that I was in the direct Cc list, I got your message in
encrypted form as well. So, erm, I'm reading it "fine" now that I
figured out how to get my MUA to decrypt it. Probably not what you want
for patch submissions, though... :P

-Toke
Alexander Lobakin May 3, 2022, 9:17 p.m. UTC | #5 
From: Alexei Starovoitov <alexei.starovoitov@gmail.com>
Date: Wed, 20 Apr 2022 17:40:34 -0700

> On Wed, Apr 20, 2022 at 5:38 PM Alexander Lobakin <alobakin@pm.me> wrote:
>
> Again?
>
> -----BEGIN PGP MESSAGE-----
> Version: ProtonMail
>
> wcFMA165ASBBe6s8AQ/8C9y4TqXgASA5xBT7UIf2GyTQRjKWcy/6kT1dkjkF
> FldAOhehhgLYjLJzNAIkecOQfz/XNapW3GdrQDq11pq9Bzs1SJJekGXlHVIW

ProtonMail support:

"
The reason that some of the recipients are receiving PGP-encrypted
emails is that kernel.org is providing public keys for those
recipients (ast@kernel.org and toke@kernel.org specifically) via WKD
(Web Key Directory), and our API automatically encrypts messages
when a key is served over WKD.

Unfortunately, there is currently no way to disable encryption for
recipients that server keys over WKD but the recipients should be
able to decrypt the messages using the secret keys that correspond
to their public keys provided by kernel.org.
This is applicable both to messages sent via the ProtonMail web app,
and messages sent via Bridge app.

We have forwarded your feedback to the appropriate teams, and we
will see if we can implement a disable encryption option for these
cases. Unfortunately, we cannot speculate when we might implement
such an option.
"

Weeeeeird, it wasn't like that a year ago.
Anyway, since it's address specific and for now I observed this only
for ast@ and toke@, can I maybe send the series adding your Gmail
account rather that korg one? Alternatively, I can send it from my
Intel address if you prefer (thankfully, it doesn't encrypt anything
without asking), I just didn't want to mix personal stuff with corp.

>
> Sorry I'm tossing the series out of patchwork.

Thanks,
Al
Toke Høiland-Jørgensen May 4, 2022, 11:34 a.m. UTC | #6 
Alexander Lobakin <alobakin@pm.me> writes:

> From: Alexei Starovoitov <alexei.starovoitov@gmail.com>
> Date: Wed, 20 Apr 2022 17:40:34 -0700
>
>> On Wed, Apr 20, 2022 at 5:38 PM Alexander Lobakin <alobakin@pm.me> wrote:
>>
>> Again?
>>
>> -----BEGIN PGP MESSAGE-----
>> Version: ProtonMail
>>
>> wcFMA165ASBBe6s8AQ/8C9y4TqXgASA5xBT7UIf2GyTQRjKWcy/6kT1dkjkF
>> FldAOhehhgLYjLJzNAIkecOQfz/XNapW3GdrQDq11pq9Bzs1SJJekGXlHVIW
>
> ProtonMail support:
>
> "
> The reason that some of the recipients are receiving PGP-encrypted
> emails is that kernel.org is providing public keys for those
> recipients (ast@kernel.org and toke@kernel.org specifically) via WKD
> (Web Key Directory), and our API automatically encrypts messages
> when a key is served over WKD.
>
> Unfortunately, there is currently no way to disable encryption for
> recipients that server keys over WKD but the recipients should be
> able to decrypt the messages using the secret keys that correspond
> to their public keys provided by kernel.org.
> This is applicable both to messages sent via the ProtonMail web app,
> and messages sent via Bridge app.
>
> We have forwarded your feedback to the appropriate teams, and we
> will see if we can implement a disable encryption option for these
> cases. Unfortunately, we cannot speculate when we might implement
> such an option.
> "
>
> Weeeeeird, it wasn't like that a year ago.

Well, they're also doing something non-standard with their WKD
retrieval, so maybe that changed? GPG itself will refuse to retrieve a
key that doesn't have the email address specified in the key itself:

$ gpg --locate-keys toke@kernel.org
gpg: key 4A55C497F744F705: no valid user IDs
gpg: Total number processed: 1
gpg:           w/o user IDs: 1
gpg: error retrieving 'toke@kernel.org' via WKD: No fingerprint

Given that they do it this way, I suppose this will affect every
@kernel.org address that has a PGP key attached (of which there are
currently 519, according to pgpkeys.git)...

-Toke