)]}'
{"/PATCHSET_LEVEL":[{"author":{"_account_id":1001964,"name":"Erhan Kurubas","display_name":"Erhan Kurubas","email":"erhan.kurubas@espressif.com","username":"erhankur"},"change_message_id":"ada690784e659a906b4bf30c3da42124c136ae08","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":1,"id":"e5d807cd_903a9438","updated":"2023-01-21 19:34:04.000000000","message":"Antonio, in the PS-7443, you can see the use case for the new callback.","commit_id":"5e622ea4478e2c13c5406e9d222ecf95ad918d48"},{"author":{"_account_id":1001494,"name":"sborshch","username":"sborshch"},"change_message_id":"78605064f31df077f37f160642d0af41328792ce","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":1,"id":"66a02299_f4bbfa56","updated":"2023-01-22 10:39:52.000000000","message":"Non-optimal implementation: you perform check for read_stack() every time stacking-\u003eread_stack() called. You can write wrapper around target_read_buffer() and assign this wrapper to read_stack() in rtos_create() if custom read_stack() not exist or just force programmer to assign this wrapper to stacking struct if custom implementation not exist.","commit_id":"5e622ea4478e2c13c5406e9d222ecf95ad918d48"},{"author":{"_account_id":1000021,"name":"Antonio Borneo","email":"borneo.antonio@gmail.com","username":"borneoa"},"change_message_id":"d63f86be8c681a7b2d257642d6468eb98bc01365","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":1,"id":"f2ad9163_9f09e73d","in_reply_to":"66a02299_f4bbfa56","updated":"2023-01-22 15:36:39.000000000","message":"Yes, I agree with your analysis, but this low level optimization doesn\u0027t help so much with OpenOCD performance.\nCalling target_read_buffer() involves the communication through JTAG with all the other stacks involved (e.g. USB), and this is the real bottleneck.\n\nThen, depending on the architecture of the host, checking \u0027stacking-\u003eread_stack\u0027 can be quite fast, while calling a function pointer can be much heavier.\nHaving a default function assigned to \u0027stacking-\u003eread_stack\u0027, which is then always called, can be worst than checking the pointer and calling target_read_buffer() by default.\n\nAnyway all these possible implementation will not really change the performance of OpenOCD. I would prefer the approach that makes the code more readable (whatever it means) and the ones that optimizes JTAG communication.","commit_id":"5e622ea4478e2c13c5406e9d222ecf95ad918d48"}]}