Re: [PATCH net v5 3/5] net: macb: move ring size computation to functions
From: Jakub Kicinski <kuba@kernel.org>
Date: 2025-09-11 23:39:54
Also in:
linux-devicetree, lkml
On Thu, 11 Sep 2025 11:14:52 +0200 Théo Lebrun wrote:
quoted
it would be good to have these functions as inline. May be as a separate patch.I don't see why? Compilers are clever pieces, they'll know to inline it. If we added inline to macb_{tx,rx}_ring_size_per_queue(), should we also add it to macb_dma_desc_get_size()? I do not know, but my compiler decided to inline it as well. It might make other decisions on other platforms. Last point I see: those two functions are not called in the hotpath, only at alloc & free. If we talk about inline for the theoretical speed gain, then it doesn't matter in that case. If it is a code size aspect, then once again the compiler is more aware than myself. I don't like the tone, but it is part of the kernel doc and is on topic: https://www.kernel.org/doc/html/latest/process/coding-style.html#the-inline-disease
👍️ FWIW, please don't sprinkle inlines.