diff options
| author | salmonik <salmonik@tuta.io> | 2026-08-02 23:28:29 +0200 |
|---|---|---|
| committer | salmonik <salmonik@tuta.io> | 2026-08-02 23:28:29 +0200 |
| commit | f846a9551bf30d61026fa9c9ccea78479ced321b (patch) | |
| tree | 10158ca300cc51f44fc9afee9c2b3092554d880b /bins/hello | |
| parent | v0.5.7: guard reports the drop's size/align — identifies the clobbered struct (diff) | |
| download | rs-dos-f846a9551bf30d61026fa9c9ccea78479ced321b.tar.gz rs-dos-f846a9551bf30d61026fa9c9ccea78479ced321b.zip | |
v0.5.7: drop test-poison hooks; vt heap-struct verify at shell entry
The OptiPlex's first banner write #GPs when the vt cells VecData holds
firmware 0xAA fill instead of a real buffer pointer. The poison/fill
test hooks (stack + free-heap + free-page 0x00AAAAAA) were removed from
the release image — they changed boot behavior and may walk a corrupt
free list over live blocks. Added vt::debug_verify() (cells/rendered
ptr/cap/len + in-heap check) at shell entry: the next real-hardware
boot shows whether the corruption is present at shell entry or lands
during the banner write.
Diffstat (limited to 'bins/hello')
0 files changed, 0 insertions, 0 deletions