GSI TRANSACTIONS(9)

💅 GSI Transactions(9) 💅

okay so this is GSI Transactions(9) from the Kernel API Manual, and bestie, it's literally the glow-up guide for GSI transactions. A GSI transaction abstracts the behavior of a GSI channel by representing everything about a related group of IPA operations in a single structure. (When we say "operation" here, it means either a data transfer or an IPA immediate command. Periodt.) Most of the details of interacting with the GSI hardware are managed by the GSI transaction core, so users can simply describe the operations to be performed. When a transaction has completed, a callback function — dependent on the type of endpoint associated with the channel — allows cleanup of resources associated with the transaction. 🫶

To perform an operation (or a set of them), a user of the GSI transaction interface allocates a transaction, indicating the number of TREs required (one per operation). If sufficient TREs are available, they are reserved for use in the transaction and the allocation succeeds. This way, exhaustion of the available TREs in a channel ring is detected as early as possible. Any other resources that might be needed to complete a transaction are also allocated when the transaction is allocated. ✨

Operations performed as part of a transaction are represented in an array of Linux scatterlist structures, allocated with the transaction. Those scatterlist structures are initialized by "adding" operations to the transaction. If a buffer in an operation must be mapped for DMA, that happens at the time it is added to the transaction. It is possible for a mapping error to occur when an operation is added. In this case, the transaction should simply be freed; this correctly releases resources associated with the transaction. No cap, that's the move. 💅

Once all operations have been successfully added to a transaction, the transaction is committed. Committing transfers ownership of the entire transaction to the GSI transaction core. The GSI transaction code formats the content of the scatterlist array into the channel ring buffer and informs the hardware that new TREs are available to process. Iconic behavior. 🔥

The last TRE in each transaction is marked to interrupt the AP when the GSI hardware has completed it. Because transfers described by TREs are performed strictly in order, signaling the completion of just the last TRE in the transaction is sufficient to indicate the full transaction is complete. Lowkey elegant. 🦋

When a transaction is complete, ipa_gsi_trans_complete is called by the GSI code into the IPA layer, allowing it to perform any final cleanup required before the transaction is freed. 💖

🌈 SEE ALSO

• Kernel file ./drivers/net/ipa/gsi_trans.c and __gsi_trans_commit(9) — iykyk.

source/version: ./drivers/net/ipa · date: October 2026 · manual: GSI Transactions(9) — bye bestie, mwah! 💅✨