On Wed, Oct 1, 2025 at 3:18 AM Guan-Chun Wu <[email protected]> wrote: > > On Fri, Sep 26, 2025 at 04:33:12PM -0700, Caleb Sander Mateos wrote: > > On Thu, Sep 25, 2025 at 11:59 PM Guan-Chun Wu <[email protected]> > > wrote: > > > > > > From: Kuan-Wei Chiu <[email protected]> > > > > > > Replace the use of strchr() in base64_decode() with precomputed reverse > > > lookup tables for each variant. This avoids repeated string scans and > > > improves performance. Use -1 in the tables to mark invalid characters. > > > > > > Decode: > > > 64B ~1530ns -> ~75ns (~20.4x) > > > 1KB ~27726ns -> ~1165ns (~23.8x) > > > > > > Signed-off-by: Kuan-Wei Chiu <[email protected]> > > > Co-developed-by: Guan-Chun Wu <[email protected]> > > > Signed-off-by: Guan-Chun Wu <[email protected]> > > > --- > > > lib/base64.c | 66 ++++++++++++++++++++++++++++++++++++++++++++++++---- > > > 1 file changed, 61 insertions(+), 5 deletions(-) > > > > > > diff --git a/lib/base64.c b/lib/base64.c > > > index 1af557785..b20fdf168 100644 > > > --- a/lib/base64.c > > > +++ b/lib/base64.c > > > @@ -21,6 +21,63 @@ static const char base64_tables[][65] = { > > > [BASE64_IMAP] = > > > "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+,", > > > }; > > > > > > +static const s8 base64_rev_tables[][256] = { > > > + [BASE64_STD] = { > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 62, -1, > > > -1, -1, 63, > > > + 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, -1, -1, -1, > > > -1, -1, -1, > > > + -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, > > > 12, 13, 14, > > > + 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, -1, -1, > > > -1, -1, -1, > > > + -1, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, > > > 38, 39, 40, > > > + 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + }, > > > + [BASE64_URLSAFE] = { > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > 62, -1, -1, > > > + 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, -1, -1, -1, > > > -1, -1, -1, > > > + -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, > > > 12, 13, 14, > > > + 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, -1, -1, > > > -1, -1, 63, > > > + -1, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, > > > 38, 39, 40, > > > + 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + }, > > > + [BASE64_IMAP] = { > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 62, 63, > > > -1, -1, -1, > > > + 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, -1, -1, -1, > > > -1, -1, -1, > > > + -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, > > > 12, 13, 14, > > > + 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, -1, -1, > > > -1, -1, -1, > > > + -1, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, > > > 38, 39, 40, > > > + 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, > > > -1, -1, -1, > > > + }, > > > > Do we actually need 3 separate lookup tables? It looks like all 3 > > variants agree on the value of any characters they have in common. So > > we could combine them into a single lookup table that would work for a > > valid base64 string of any variant. The only downside I can see is > > that base64 strings which are invalid in some variants might no longer > > be rejected by base64_decode(). > > > > In addition to the approach David mentioned, maybe we can use a common > lookup table for A–Z, a–z, and 0–9, and then handle the variant-specific > symbols with a switch. > > For example: > > static const s8 base64_rev_common[256] = { > [0 ... 255] = -1, > ['A'] = 0, ['B'] = 1, /* ... */, ['Z'] = 25, > ['a'] = 26, /* ... */, ['z'] = 51, > ['0'] = 52, /* ... */, ['9'] = 61, > }; > > static inline int base64_rev_lookup(u8 c, enum base64_variant variant) { > s8 v = base64_rev_common[c]; > if (v != -1) > return v; > > switch (variant) { > case BASE64_STD: > if (c == '+') return 62; > if (c == '/') return 63; > break; > case BASE64_IMAP: > if (c == '+') return 62; > if (c == ',') return 63; > break; > case BASE64_URLSAFE: > if (c == '-') return 62; > if (c == '_') return 63; > break; > } > return -1; > } > > What do you think?
That adds several branches in the hot loop, at least 2 of which are unpredictable for valid base64 input of a given variant (v != -1 as well as the first c check in the applicable switch case). That seems like it would hurt performance, no? I think having 3 separate tables would be preferable to making the hot loop more branchy. Best, Caleb
