apply_svp1_workaround(9)

💅 apply_svp1_workaround(9) — from the Kernel API Manual, bestie 🎀

okay so bestie, we're in the Kernel API Manual today and this page is called apply_svp1_workaround. no cap, the whole point of it is: apply the workaround required for the scenario where svp0_height_next != 0 && svp1_height_next == 0. that's the vibe. that's the whole assignment. ✨

---

🌸 SYNOPSIS 🌸

void apply_svp1_workaround (uint32_t svp0_start_sw_idx_next , uint32_t svp0_num_swaths_next , uint32_t svp1_num_swaths_next , uint16_t viewport_start , uint16_t viewport_size , uint16_t swath_height , uint8_t access_direction , uint16_t *svp1_start_line );

like, it's void. it returns nothing. it just does things. main character energy honestly. 👑

---

🦋 ARGUMENTS 🦋

so here's the part where I have to be real with you: the manpage literally just says -- undescribed -- for every single one of these. I'm not out here inventing lore 💀 — so I'm passing the mystery straight to you, no filter:

• svp0_start_sw_idx_next — -- undescribed -- 🫥
• svp0_num_swaths_next — -- undescribed -- 🫥
• svp1_num_swaths_next — -- undescribed -- 🫥
• viewport_start — -- undescribed -- 🫥
• viewport_size — -- undescribed -- 🫥
• swath_height — -- undescribed -- 🫥
• access_direction — -- undescribed -- 🫥
• svp1_start_line — -- undescribed -- 🫥

(note that svp1_start_line is a uint16_t *, so it's a pointer — it's the one getting written into, iykyk 👀. the rest are plain values passed in.)

---

✨ DESCRIPTION ✨

the workaround requires the following, and I'm quoting the math exactly because this is the whole entire point of the function:

• svp1_start_line = svp0_start_line_next + svp0_height_next for access_direction == 0
• svp1_start_line = svp0_start_line_next - svp0_height_next for access_direction == 1

so like… one direction you add, the other direction you subtract. that's it. that's the tea. 🍓

and then, direct from the DCN IP team — this is the part where you do NOT get creative:

From DCN IP team: Do not clamp any value from this calculation, just used the calculated value in svp1_start_line as is

read that again bestie. no clamping. none. zero. you take whatever number falls out of the calculation and you put it in svp1_start_line as is. do not "fix" it, do not "guard" it, do not get cute with it. the IP team said what they said 💅

and here's that giant asterisk banner from the original, preserved in all its chaotic glory because I refuse to edit the source:

******************************************************************************************************************

slay. moving on. 😌

---

🔥 SEE ALSO 🔥

Kernel file:

./drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_standalone_libraries/alternate_pstate_shared_lib.c

and the whole friend group: 🫶

• in_circular_range(9)
• calculate_start_line_and_height_from_swath(9)
• get_swath_deadlines(9)
• calculate_swath_deadline_dst(9)
• is_dst_in_next_frame(9)
• is_end_swath_in_next_frame(9)
• is_svp_dst_in_next_frame(9)
• calculate_swath_idx_from_dst(9)
• populate_svp_params(9)
• populate_start_index_num_swaths_helper(9)
• calculate_start_index_num_swaths(9)
• calculate_hubp_start_end_lines(9)
• get_prefetch_end_line(9)
• get_effective_vblank_start(9)

---

⭐ little sign-off moment: this page comes to you from ./drivers/gpu/drm/amd/display/..., dated October 2026. screenshot it, frame it, put it on your wall. mwah. 💖🧚🌈