Presumably i386 needs the same fixup as well?
- Queries
- All Stories
- Search
- Advanced Search
- Transactions
- Transaction Logs
Advanced Search
Yesterday
Tue, Mar 26
Mon, Mar 25
Sat, Mar 23
Thu, Mar 21
Wed, Mar 20
Tue, Mar 19
I ran into this testing the DDP changes for NFS (was ensuring that TLS offload was used instead of DDP for supported TLS 1.2 ciphers). This race was triggered on the server side where presumably the client had sent an RPC immediately after completing the TLS handshake.
Rework enabling
Fix various bugs from testing
I really wish we could transition KASSERT gracefully to using __VA_ARGS__ and away from the nested ()'s, but I'm not sure of a good way to permit both syntaxes concurrently.
Oh, hmm, I would think of topology differently TBH. That is, my hope/expectation was to get a hierarchical view similar to one of the views you can get with lspci where you print the PCI devices as an indented hierarchy similar to the output from devinfo, where devices are children (and indented) under their parent PCI-PCI bridges, etc. We also tend to use "domain" instead of "node" for NUMA in general in prose descriptions.
Modulo dropping support for the count argument from kdb_cpu_clr_breakpoint.
Typo in commit log: s/excpeted/expected/
I'd just skip to removing outright I think.