💬 fanquake commented on pull request "sanitizer: symbolizer improvements":
(https://github.com/bitcoin/bitcoin/pull/28814#discussion_r1385093144)
Yea, the worst that can happen is the problem we already have. Let's see if we can make this more generic.

💬 fanquake commented on pull request "sanitizer: symbolizer improvements":
(https://github.com/bitcoin/bitcoin/pull/28814#discussion_r1385106168)
I guess there are also other issues here, as the ASAN job does find the symbolizer, but not at that path:
```bash
AddressSanitizer: reading suppressions file at /ci_container_base/test/sanitizer_suppressions/ubsan
==28779==Using llvm-symbolizer found at: /usr/bin/llvm-symbolizer-17
==28779==AddressSanitizer Init done
Error parsing command line arguments: ==28779==__tls_get_addr: DTLS_Find 0x7f7962720770 2
```

👍 laanwj approved a pull request: "Add support for RNDR/RNDRRS for AArch64 on Linux"
(https://github.com/bitcoin/bitcoin/pull/26839#pullrequestreview-1718023531)
Tested ACK aee5404e02e203a256c1a97b629b9b107cc8bb07
I have verified this PR on the only available hardware (Amazon Graviton 3, m7g.medium instance). I was hoping for real hardware to become available but it looks like the recent crop of SoCs (such as RPI5) still doesn't have support for this extension.
In any case I've checked that the capability is detected correctly, and that RNDRRS is called four times at startup and RNDR frequently during runtime, and that the returned values look random.
...

💬 maflcko commented on pull request "sanitizer: symbolizer improvements":
(https://github.com/bitcoin/bitcoin/pull/28814#discussion_r1385139952)
Yes, the task does not have llvm installed, but only `llvm-17`

💬 fanquake commented on pull request "sanitizer: symbolizer improvements":
(https://github.com/bitcoin/bitcoin/pull/28814#discussion_r1385146001)
Yea, my question is why is it now failing, if it's successfully finding a symbolizer.

💬 fanquake commented on pull request "sanitizer: symbolizer improvements":
(https://github.com/bitcoin/bitcoin/pull/28814#discussion_r1385151899)
Ah, it looks like the `verbosity=2` output is breaking things that are parsing output:
```bash
/usr/bin/python3.11 ../test/util/test_runner.py
2023-11-07 15:13:31,154 - ERROR - Error mismatch:
Expected: Error parsing command line arguments: Invalid command 'foo'
```
and the ASAN job here would otherwise be working as expected.

💬 mzumsande commented on issue "v26.0 Testing":
(https://github.com/bitcoin/bitcoin/issues/28718#issuecomment-1799048645)
The link to the release notes draft doesn't work for me.

💬 fanquake commented on issue "v26.0 Testing":
(https://github.com/bitcoin/bitcoin/issues/28718#issuecomment-1799053374)
Should be fixed now.

💬 achow101 commented on pull request "wallet: prevent bugs from invalid transaction heights with asserts, comments, and refactoring":
(https://github.com/bitcoin/bitcoin/pull/28546#issuecomment-1799093337)
ACK f06016d77d848133fc6568f287bb86b644c9fa69

🚀 achow101 merged a pull request: "wallet: prevent bugs from invalid transaction heights with asserts, comments, and refactoring"
(https://github.com/bitcoin/bitcoin/pull/28546)

💬 TheCharlatan commented on pull request "guix: switch to 6.1 kernel headers over 5.15":
(https://github.com/bitcoin/bitcoin/pull/28786#issuecomment-1799123554)
Guix builds (aarch64)
```
find guix-build-$(git rev-parse --short=12 HEAD)/output/ -type f -print0 | env LC_ALL=C sort -z | xargs -r0 sha256sum
9aba5e0ea27cbf609baac52e65f0eb63bc42aa7e1865108422d30b0bad01a7da guix-build-380e3655631b/output/aarch64-linux-gnu/SHA256SUMS.part
67644dd783c781ba6ae6bffa91204d3add423fe856b3aac09cb18c5f6db18599 guix-build-380e3655631b/output/aarch64-linux-gnu/bitcoin-380e3655631b-aarch64-linux-gnu-debug.tar.gz
0130236c032fbdcb5b791510d2ecf61239b434a1996b4b35ad7e7
...

📝 maflcko opened a pull request: "fuzz: Avoid timeout and bloat in bloom_filter target"
(https://github.com/bitcoin/bitcoin/pull/28815)
If the fuzz input contains invalid data, abort early. This will teach the fuzz engine to look for useful data and avoids bloating the fuzz input folder with useless data.

👍 TheCharlatan approved a pull request: "guix: switch to 6.1 kernel headers over 5.15"
(https://github.com/bitcoin/bitcoin/pull/28786#pullrequestreview-1718170494)
ACK 380e3655631b2604235fb971f5a92444d43d554b

Agree that this should just be done implicitly with a time machine bump.

💬 maflcko commented on issue "fuzz: Fix timeouts":
(https://github.com/bitcoin/bitcoin/issues/28812#issuecomment-1799137527)
Bloom fixed in https://github.com/bitcoin/bitcoin/pull/28815

💬 maflcko commented on pull request "fuzz: Avoid timeout and bloat in bloom_filter target":
(https://github.com/bitcoin/bitcoin/pull/28815#issuecomment-1799139133)
Can be tested by running https://github.com/bitcoin/bitcoin/files/13280645/clusterfuzz-testcase-bloom_filter-5968259782148096.dmp.not.txt and looking at the runtime

🤔 glozow reviewed a pull request: "Fee Estimator updates from Validation Interface/CScheduler thread"
(https://github.com/bitcoin/bitcoin/pull/28368#pullrequestreview-1718119056)
Maybe add a `SyncWithValidationInterfaceQueue` to the start of the fee estimation RPCs? To avoid spurious failures in `feature_fee_estimation.py`. Also, I can imagine scripts that grab fee estimates once per block as soon as the block arrives - should probably make sure they're fresh.

💬 glozow commented on pull request "Fee Estimator updates from Validation Interface/CScheduler thread":
(https://github.com/bitcoin/bitcoin/pull/28368#discussion_r1385183779)
09436b21e84cc6cd979fe4942ba1c415c4bc91be can be squashed

💬 glozow commented on pull request "Fee Estimator updates from Validation Interface/CScheduler thread":
(https://github.com/bitcoin/bitcoin/pull/28368#discussion_r1385215139)
ac82f58e45421aa5aec1d3e6e992d1c369472a10

```suggestion
const auto& hash = tx_info.m_tx->GetHash();
```

💬 glozow commented on pull request "Fee Estimator updates from Validation Interface/CScheduler thread":
(https://github.com/bitcoin/bitcoin/pull/28368#discussion_r1385207240)
6f40c32fa8dfc742cf5ca7daa28615e066872aec

maybe stop making a copy, and let it be a `Txid`

```suggestion
const auto& hash = it->GetTx().GetHash();
```

💬 glozow commented on pull request "Fee Estimator updates from Validation Interface/CScheduler thread":
(https://github.com/bitcoin/bitcoin/pull/28368#discussion_r1385219874)
you're changing how the fuzz inputs are processed here - maybe have `m_has_no_mempool_parents = ConsumeBool()` and make the others = false/false/true?