[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20241029182827.u4az53onywedeot6@skbuf>
Date: Tue, 29 Oct 2024 20:28:27 +0200
From: Vladimir Oltean <olteanv@...il.com>
To: Jacob Keller <jacob.e.keller@...el.com>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
Eric Dumazet <edumazet@...gle.com>,
Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>,
Tony Nguyen <anthony.l.nguyen@...el.com>,
Przemek Kitszel <przemyslaw.kitszel@...el.com>,
Masahiro Yamada <masahiroy@...nel.org>,
linux-kernel@...r.kernel.org, netdev@...r.kernel.org,
Vladimir Oltean <vladimir.oltean@....com>
Subject: Re: [PATCH net-next v2 0/9] lib: packing: introduce and use
(un)pack_fields
On Fri, Oct 25, 2024 at 05:04:52PM -0700, Jacob Keller wrote:
> This series improves the packing library with a new API for packing or
> unpacking a large number of fields at once with minimal code footprint. The
> API is then used to replace bespoke packing logic in the ice driver,
> preparing it to handle unpacking in the future. Finally, the ice driver has
> a few other cleanups related to the packing logic.
>
> The pack_fields and unpack_fields functions have the following improvements
> over the existing pack() and unpack() API:
>
> 1. Packing or unpacking a large number of fields takes significantly less
> code. This significantly reduces the .text size for an increase in the
> .data size which is much smaller.
>
> 2. The unpacked data can be stored in sizes smaller than u64 variables.
> This reduces the storage requirement both for runtime data structures,
> and for the rodata defining the fields. This scales with the number of
> fields used.
>
> 3. Most of the error checking is done at compile time, rather than
> runtime via CHECK_PACKED_FIELD_* macros. This saves wasted computation
> time, *and* catches errors in the field definitions immediately instead
> of only after the offending code executes.
>
> The actual packing and unpacking code still uses the u64 size
> variables. However, these are converted to the appropriate field sizes when
> storing or reading the data from the buffer.
>
> One complexity is that the CHECK_PACKED_FIELD_* macros need to be defined
> one per size of the packed_fields array. This is because we don't have a
> good way to handle the ordering checks otherwise. The C pre-processor is
> unable to generate and run variable length loops at compile time.
>
> This is a significant amount of macro code, ~22,000 lines of code. To
> ensure it is correct and to avoid needing to store this directly in the
> kernel history, this file is generated as <generated/packing-checks.h> via
> a small C program, gen_packing_checks. To generate this, we need to update
> the top level Kbuild process to include the compilation of
> gen_packing_checks and execution to generate the packing-checks.h file.
>
> Signed-off-by: Jacob Keller <jacob.e.keller@...el.com>
> ---
> Changes in v2:
> - Add my missing sign-off to the first patch
> - Update the descriptions for a few patches
> - Only generate CHECK_PACKED_FIELDS_N when another module selects it
> - Add a new patch introducing wrapper structures for the packed Tx and Rx
> queue context, suggested by Vladimir.
> - Drop the now unnecessary macros in ice, thanks to the new types
> - Link to v1: https://lore.kernel.org/r/20241011-packing-pack-fields-and-ice-implementation-v1-0-d9b1f7500740@intel.com
For the set:
Reviewed-by: Vladimir Oltean <olteanv@...il.com>
Thanks for working on this!
Powered by blists - more mailing lists