vgsp: CONT-97 - gsp_phys_ds_spy re-clears DS_REPORT_MASK/_2 every sample + maskholds/last_mask logging (W1C confirmed: pgraph-rw 0x405840<-0x20 reads back 00000000, no synthetic calibration possible); slots NAMED: c597 SetPipeline[j] only j=0..5, window binds j=1/2/5 ProgramAddressA=0x19a0900/0x19a0200/0x19a0000 byte-equal SPH_INFO stage VAs; DS_HWW_ESR bits 0..23 = SPH0..23 checks (RM DS_SPH<n>_ERR=0x6e+n) so j=9/10/11 binds cannot exist - {5,9,10,11} are per-SPH validation checks; pgraph-rw A/B: golden PATCH DS writes are TGA constraint logic ALREADY LIVE (0x700/0x800) not SPH-related; DS-spy captures: hits all badf1201 poison or zero (timing anomaly recorded); GATE1 black 3 boots ds=a4000e20 sph=0,019a0900,019a0200 0/900 pages; guest root capture 99.92% black; next: beef window's only call form is op=4 IMMD_DATA_METHOD on CallMme[0].Macro 0x3800 (MME-WINRUN off=a0 op=4 cnt=48 mthd=3800) never executed by MME (pcv=0; held windows retire, call-bearing park) while prelude resident - decode run vs MME call spec offline (mme_replay_slots81.txt) + re-emit client's own call shape (trigger 0x3800 + params 0x3804); if op=4 correct run pure-method window test keep_mask=0+hold=N to decide if replay needs MME at all - SESSION 153 r1326-r1327

0 files changed, 0 insertions, 0 deletions