crypto/internal/bigmod: flag amd64 assembly as noescape

[FiloSottile]

I had overlooked it, causing a performance loss.

name                    old time/op    new time/op    delta
DecryptPKCS1v15/2048-8    1.50ms ± 0%    1.48ms ± 0%   -0.95%  (p=0.000 n=9+10)
DecryptPKCS1v15/3072-8    4.64ms ± 1%    4.60ms ± 0%   -0.82%  (p=0.000 n=8+10)
DecryptPKCS1v15/4096-8    10.7ms ± 0%    10.6ms ± 1%   -0.99%  (p=0.000 n=10+10)
EncryptPKCS1v15/2048-8     158µs ± 0%     157µs ± 0%   -0.63%  (p=0.000 n=10+10)
DecryptOAEP/2048-8        1.50ms ± 0%    1.48ms ± 0%   -1.09%  (p=0.000 n=9+10)
EncryptOAEP/2048-8         161µs ± 0%     160µs ± 0%   -0.34%  (p=0.000 n=9+10)
SignPKCS1v15/2048-8       1.55ms ± 0%    1.53ms ± 1%   -1.32%  (p=0.000 n=10+10)
VerifyPKCS1v15/2048-8      157µs ± 0%     157µs ± 0%   -0.33%  (p=0.004 n=9+10)
SignPSS/2048-8            1.55ms ± 0%    1.54ms ± 0%   -1.14%  (p=0.000 n=10+10)
VerifyPSS/2048-8           160µs ± 0%     160µs ± 0%   -0.32%  (p=0.000 n=10+10)

name                    old alloc/op   new alloc/op   delta
DecryptPKCS1v15/2048-8    15.0kB ± 0%     0.6kB ± 0%  -95.74%  (p=0.000 n=10+10)
DecryptPKCS1v15/3072-8    17.9kB ± 0%     3.5kB ± 0%  -80.65%  (p=0.000 n=10+10)
DecryptPKCS1v15/4096-8    19.1kB ± 0%     4.7kB ± 0%  -75.25%  (p=0.000 n=10+10)
EncryptPKCS1v15/2048-8    7.51kB ± 0%    1.17kB ± 0%  -84.39%  (p=0.000 n=10+10)
DecryptOAEP/2048-8        15.3kB ± 0%     0.9kB ± 0%  -94.29%  (p=0.000 n=10+10)
EncryptOAEP/2048-8        7.74kB ± 0%    1.40kB ± 0%  -81.86%  (p=0.000 n=10+10)
SignPKCS1v15/2048-8       21.6kB ± 0%     0.9kB ± 0%  -95.86%  (p=0.000 n=10+10)
VerifyPKCS1v15/2048-8     7.25kB ± 0%    0.91kB ± 0%  -87.42%  (p=0.000 n=10+10)
SignPSS/2048-8            22.0kB ± 0%     1.3kB ± 0%  -94.12%  (p=0.000 n=10+10)
VerifyPSS/2048-8          7.46kB ± 0%    1.12kB ± 0%  -84.98%  (p=0.000 n=10+10)

name                    old allocs/op  new allocs/op  delta
DecryptPKCS1v15/2048-8      54.0 ± 0%       4.0 ± 0%  -92.59%  (p=0.000 n=10+10)
DecryptPKCS1v15/3072-8      60.0 ± 0%      10.0 ± 0%  -83.33%  (p=0.000 n=10+10)
DecryptPKCS1v15/4096-8      60.0 ± 0%      10.0 ± 0%  -83.33%  (p=0.000 n=10+10)
EncryptPKCS1v15/2048-8      29.0 ± 0%       7.0 ± 0%  -75.86%  (p=0.000 n=10+10)
DecryptOAEP/2048-8          60.0 ± 0%      10.0 ± 0%  -83.33%  (p=0.000 n=10+10)
EncryptOAEP/2048-8          35.0 ± 0%      13.0 ± 0%  -62.86%  (p=0.000 n=10+10)
SignPKCS1v15/2048-8         77.0 ± 0%       5.0 ± 0%  -93.51%  (p=0.000 n=10+10)
VerifyPKCS1v15/2048-8       28.0 ± 0%       6.0 ± 0%  -78.57%  (p=0.000 n=10+10)
SignPSS/2048-8              82.0 ± 0%      10.0 ± 0%  -87.80%  (p=0.000 n=10+10)
VerifyPSS/2048-8            33.0 ± 0%      11.0 ± 0%  -66.67%  (p=0.000 n=10+10)

[gopherbot]

Change https://go.dev/cl/460542 mentions this issue: crypto/internal/bigmod: flag amd64 assembly as noescape

[FiloSottile]

@gopherbot please open a Go 1.20 backport. This is arguably partially mitigating a performance regression in Go 1.20, it's a "bug" in new code introduced in Go 1.20, and the patch is as safe as it gets.

[gopherbot]

Backport issue(s) opened: #58505 (for 1.20).

Remember to create the cherry-pick CL(s) as soon as the patch is submitted to master, according to https://go.dev/wiki/MinorReleases.

[gopherbot]

Change https://go.dev/cl/471855 mentions this issue: [release-branch.go1.20] crypto/internal/bigmod: flag amd64 assembly as noescape