zero-copy slot hashes sysvar (with checked alternatives) #152
Conversation
Update from upstream
…topian/pinocchio into rustopian/slot-hashes-sysvar
…topian/pinocchio into rustopian/slot-hashes-sysvar
| // Reject oversized entry counts to prevent surprises. | ||
| if num_entries > MAX_ENTRIES { | ||
| return Err(ProgramError::InvalidArgument); | ||
| } | ||
|
|
||
| let required_len = NUM_ENTRIES_SIZE + num_entries * ENTRY_SIZE; | ||
| if buffer.len() < required_len { | ||
| return Err(ProgramError::InvalidArgument); | ||
| } |
There was a problem hiding this comment.
I am not sure the need for this checks. I would assume that the syscall did the right thing, so num_entries would never be greater than MAX_ENTRIES and if the length was invalid, the syscall would fail.
There was a problem hiding this comment.
removed num_entries > MAX_ENTRIES check in 551b0f9
There was a problem hiding this comment.
Do we need the if buffer.len() < required_le check? This is after the syscall was called, so if the buffer length was invalid, the syscall would have failed.
There was a problem hiding this comment.
Removed. More importantly, as I double-checked the context here, I saw that cases involving offset would happily return an incorrect number of entries to the caller (reading slot number data as a num_entries header).
fetch_into and validate_buffer_size (now named validate_and_get_buffer_capacity) have been rewritten to fix this, plus some test updates and a new test test_fetch_into_offset_avoids_incorrect_entry_count
febo
left a comment
febo
left a comment
There was a problem hiding this comment.
Looks great, just very minor things.
Co-authored-by: Fernando Otero <[email protected]>
Co-authored-by: Fernando Otero <[email protected]>
Update from upstream
rustopian
force-pushed
the
rustopian/slot-hashes-sysvar
branch
from
12ce68d to
99eb7da
Compare
Minor things addressed (and hunspell satisfied). |
rustopian
force-pushed
the
rustopian/slot-hashes-sysvar
branch
from
6b6aca4 to
6ed4829
Compare
rustopian
force-pushed
the
rustopian/slot-hashes-sysvar
branch
from
6ed4829 to
1d9cab0
Compare
|
Still hoping this will be merged... It looks done, does @febo need to review this one last time before merging? |
Got your wish, it's in! (thanks febo) |
4 participants
Adds SlotHashes sysvar support. Checked and unchecked zero-copy paths are both included. Intended to replace 121.
Binary search is used, but zerocopy
Iteratorand copyfetch/fetch_into/fetch_into_uncheckedare also implemented for those with different SlotHashes access needs.