RISC-V getrandom vDSO Ready Ahead Of Linux 6.16 With Exciting Performance
Merged last year in Linux 6.11 was getrandom() support in the vDSO for x86/x86_64 and then in Linux 6.12 was extended to LoongArch and ARM64. With the upcoming Linux 6.16 cycle, this support for faster while still secure RNG for user-space is set to come to RISC-V.
Queued within the RISC-V kernel's "for-next" Git branch is wiring up getrandom() vDSO support for this CPU architecture. The queued up code gets the cDSO implementation working on RISC-V processors.
The benchmark results shown with the pending commit are quite compelling:
Barring any last minute issues from being uncovered in this code, the RISC-V for-next patches will then be sent in for the Linux 6.16 merge window once it opens around the end of May.
Queued within the RISC-V kernel's "for-next" Git branch is wiring up getrandom() vDSO support for this CPU architecture. The queued up code gets the cDSO implementation working on RISC-V processors.
The benchmark results shown with the pending commit are quite compelling:
👁 RISC-V getrandom vDSO comparison benchmarks
Barring any last minute issues from being uncovered in this code, the RISC-V for-next patches will then be sent in for the Linux 6.16 merge window once it opens around the end of May.