Ameer Hamza
    • Create new note
    • Create a note from template
      • Sharing URL Link copied
      • /edit
      • View mode
        • Edit mode
        • View mode
        • Book mode
        • Slide mode
        Edit mode View mode Book mode Slide mode
      • Customize slides
      • Note Permission
      • Read
        • Only me
        • Signed-in users
        • Everyone
        Only me Signed-in users Everyone
      • Write
        • Only me
        • Signed-in users
        • Everyone
        Only me Signed-in users Everyone
      • Engagement control Commenting, Suggest edit, Emoji Reply
    • Invite by email
      Invitee

      This note has no invitees

    • Publish Note

      Share your work with the world Congratulations! 🎉 Your note is out in the world Publish Note

      Your note will be visible on your profile and discoverable by anyone.
      Your note is now live.
      This note is visible on your profile and discoverable online.
      Everyone on the web can find and read all notes of this public team.
      See published notes
      Unpublish note
      Please check the box to agree to the Community Guidelines.
      View profile
    • Commenting
      Permission
      Disabled Forbidden Owners Signed-in users Everyone
    • Enable
    • Permission
      • Forbidden
      • Owners
      • Signed-in users
      • Everyone
    • Suggest edit
      Permission
      Disabled Forbidden Owners Signed-in users Everyone
    • Enable
    • Permission
      • Forbidden
      • Owners
      • Signed-in users
    • Emoji Reply
    • Enable
    • Versions and GitHub Sync
    • Note settings
    • Note Insights New
    • Engagement control
    • Make a copy
    • Transfer ownership
    • Delete this note
    • Save as template
    • Insert from template
    • Import from
      • Dropbox
      • Google Drive
      • Gist
      • Clipboard
    • Export to
      • Dropbox
      • Google Drive
      • Gist
    • Download
      • Markdown
      • HTML
      • Raw HTML
Menu Note settings Note Insights Versions and GitHub Sync Sharing URL Create Help
Create Create new note Create a note from template
Menu
Options
Engagement control Make a copy Transfer ownership Delete this note
Import from
Dropbox Google Drive Gist Clipboard
Export to
Dropbox Google Drive Gist
Download
Markdown HTML Raw HTML
Back
Sharing URL Link copied
/edit
View mode
  • Edit mode
  • View mode
  • Book mode
  • Slide mode
Edit mode View mode Book mode Slide mode
Customize slides
Note Permission
Read
Only me
  • Only me
  • Signed-in users
  • Everyone
Only me Signed-in users Everyone
Write
Only me
  • Only me
  • Signed-in users
  • Everyone
Only me Signed-in users Everyone
Engagement control Commenting, Suggest edit, Emoji Reply
  • Invite by email
    Invitee

    This note has no invitees

  • Publish Note

    Share your work with the world Congratulations! 🎉 Your note is out in the world Publish Note

    Your note will be visible on your profile and discoverable by anyone.
    Your note is now live.
    This note is visible on your profile and discoverable online.
    Everyone on the web can find and read all notes of this public team.
    See published notes
    Unpublish note
    Please check the box to agree to the Community Guidelines.
    View profile
    Engagement control
    Commenting
    Permission
    Disabled Forbidden Owners Signed-in users Everyone
    Enable
    Permission
    • Forbidden
    • Owners
    • Signed-in users
    • Everyone
    Suggest edit
    Permission
    Disabled Forbidden Owners Signed-in users Everyone
    Enable
    Permission
    • Forbidden
    • Owners
    • Signed-in users
    Emoji Reply
    Enable
    Import from Dropbox Google Drive Gist Clipboard
       Owned this note    Owned this note      
    Published Linked with GitHub
    • Any changes
      Be notified of any changes
    • Mention me
      Be notified of mention me
    • Unsubscribe
    ###### tags: `ebpf` # offset type system inference algorithm ``` /* language supported (so far): reg ::= unsigned int stmt ::= reg = *(type *)(reg aop const) | *(type *)(reg aop const) = reg | reg = reg aop val | assume reg rop reg; type ::= u64 | u32 | u16 | u8 aop ::= + | - rop ::= > | <= val ::= reg | const */ ``` ## Constants ``` int STACK_BEGIN = 0; int CTX_BEGIN = 0; int PACKET_BEGIN = 0; int SHARED_BEGIN = 0; int PACKET_END = -4100; int PACKET_META = -1; int MAX_PACKET_SIZE = 0xffff; ``` ## Given ``` // assuming packet_begin, packet_end, packet_meta locations in context and context size have been given extern int PACKET_BEGIN_LOC; extern int PACKET_END_LOC; extern int PACKET_META_LOC; extern int CTX_SIZE; ``` ## Definitions ``` using region_t = enum {T_PACKET, T_CTX, T_SHARED, T_STACK}; using ptr_no_off_t = struct { region_t m_r; }; using ptr_with_offset_t = struct { region_t m_r; int m_offset; }; using interval_t = std::pair<int, int>; using weight_t = interval_t; // should be interval_t | symbol_t using slack_var_t = int; using rop_t = enum {R_GT, R_GE, R_LT, R_LE}; using dist_t = struct { slack_var_t m_slack; weight_t m_dist; }; // if m_dist is +ve, represents `begin+m_dist+s;` where s is some slack variable represented by a non-negative number m_slack, // if m_dist is -ve, represents `end+m_dist+1` where m_slack is -1 using dist_equality_t = std::pair<dist_t, dist_t> // represents constraint `p[0] = p[1];`, e.g., `begin+8+s = end` using inequality_t = struct { slack_var_t m_slack; rop_t m_rel; weight_t m_value; }; // represents `s m_rel m_value;` constraint, where e.g., `s >= [0,0]` and represents a contraint corresponding to dist_t constraint using packet_constraint_t = std::pair<dist_equality_t, inequality_t>; // actual implementation // using live_registers_t = std::array<reg_with_loc_t, 11>; // using global_offset_env_t = std::unordered_map<reg_with_loc_t, dist_t>; // using registers_state_t = struct { // live_registers_t m_cur_def; // std::shared_ptr<global_offset_env_t> m_offset_env; // } using registers_state_t = std::array<dist_t, 11>; // represents `r{n} = dist : dist_t;`, where {n} \belongs [0,10], e.g., `r1 = begin+8` using stack_state_t = std::unordered_map<int, dist_t>; // represents `sp[n] = dist;`, where n \belongs [0,511], e.g., `sp[508] = begin+16` using extra_constraints_t = struct { packet_constraint_t m_meta_and_begin; packet_constraint_t m_begin_and_end; } using ctx_dists_t = std::unordered_map<unsigned int, dist_t>; // represents `cp[n] = dist;` using ctx_t = struct { ctx_dists_t m_dists; int m_size; } ``` ## State ``` using state_t = struct { register_state_t m_reg_state; stack_state_t m_stack_state; extra_constraints_t m_extra_constraints; ctx_t m_ctx_state; slack_var_t m_slack = 0; }; ``` ## Initializations ``` std::unordered_map<cfg_node, state_t> states; states[0].m_ctx_state.m_dists = ctx_t{ PACKET_BEGIN_LOC: dist_t{slack: null, dist: 0}, PACKET_META_LOC: dist_t{slack: null, dist: -1}, PACKET_END_LOC: dist_t{slack: null, dist: -4100} }; states[0].m_ctx_state.m_size = CTX_SIZE; ``` ## Algorithm ``` def equal_dists(dist_t dist1, dist_t dist2): if dist1.slack is null and dist2.slack is null: // case for registers and stack slots if (dist1.dist >= 0 and dist2.dist >= 0) or (dist1.dist < 0 and dist2.dist < 0): return dist1.dist == dist2.dist; else: // when one register/stack slot is defined as forward distance and other with backward; not common case but might happen else if dist1.slack is not null and dist2.slack is not null: // we do not get to this branch for registers and stack slots return false; else: // when one distance contains slack and other doesn't; very likely we just need to return not equal but might have to do other analysis return false; def operator||(stack_dists_t stack_dists1, stack_dists_t stack_dists2): stack_dists_t stack_dists_joined; // bottom for (stack_slot, stack_slot_dist) in st1.m_stack_dists: if stack_slot in st2.m_stack_dists and equal_dists(stack_slot_dist, st2.m_stack_dists[stack_slot]): stack_dists_joined[stack_slot] = stack_slot_dist; return stack_dists_joined; def operator||(registers_state_t reg_state1, registers_state_t reg_state2): register_dists_t reg_dists_joined; // bottom for (reg, reg_dist) in st1.m_reg_dists: if reg in st2.m_reg_dists and equal_dists(reg_dist, st2.m_reg_dists[reg]): reg_dists_joined[reg] = reg_dist; return reg_dists_joined; def operator||(extra_constraints_t extra_cnstrnts1, extra_constraints_t extra_cnstrnts2): // represents begin+n+s = end+m; forward_and_backward_eq_t eq1 = extra_cnstrnts1[0]; forward_and_backward_eq_t eq2 = extra_cnstrnts2[0]; // s >= p inequality_t ineq1 = extra_cnstrnts1[1]; inequality_t ineq2 = extra_cnstrnts2[1]; extra_constraints_t cnstrnts_joined; // bottom // normalize end+m by moving m to lhs, begin+n-m+s = end; for both systems; eq1[0].dist = eq1[0].dist-eq1[1].dist-1; eq1[1].dist = -1; eq2[0].dist = eq2[0].dist-eq2[1].dist-1; eq2[1].dist = -1; // normalize s >= p by adding p to lhs of equality, begin+n-m+p+s = end, s >= 0; eq1[0].dist = eq1[0].dist+ineq1.value; ineq1.dist = 0; eq2[0].dist = eq2[0].dist+ineq2.value; ineq2.dist = 0; // given two normalized systems: // 1. begin+a+s = end, s >= 0 // 2. begin+b+t = end, t >= 0 // should be true that both equalities contain slack if eq1[0].slack is not null and eq2[0].slack is not null: slack_var_t s; if eq1[0].slack == eq2[0].slack: // s == t s = eq1[0].slack; else: s = s+string(global_slack_counter++); // new slack variable u // begin+min(a,b)+u = end dist_t f = dist_t{slack: s, dist: min(eq1[0].dist, eq2[0].dist)}; dist_t b = eq1[1]; forward_and_backward_eq_t eq = std::make_pair(f, b); // s >= 0 inequality_t ineq = inequality_t{slack: s, rel: ineq1.rel, value: 0}; cnstrnts_joined = std::make_pair(eq, ineq); return cnstrnts_joined; def operator||(state_t st1, state_t st2): state_t st_joined = _|_; // bottom st_joined.m_reg_state = st1.m_reg_state || st2.m_reg_state; st_joined.m_stack_state = st1.m_stack_state || st2.m_stack_state; st_joined.m_extra_constraints = st1.m_extra_constraints || st2.m_extra_constraints; st_joined.m_slack = std::max(st1.m_slack, st2.m_slack); return st_joined; def join(cfg_node bb): state_t state = _|_; // state is bottom cfg_node[] all_preds = cfg.get_all_predecessors(bb); for i in [0..all_preds.length()]: if i == 0: state = states[all_preds[i]]; else: state = state || states[all_preds[i]]; return state; ``` ``` for bb in cfg.topological_sort(): unsigned int global_slack_counter = 0; if cfg.num_of_predecessors(bb) == 1: state = states[cfg.get_predecessor(bb)]; else if cfg.num_of_predecessors(bb) > 1: state = join_at(bb); for instr in bb: match(instr) with: ``` ``` va = *(T *)(vb arithm_oprtr vc) -> ptr_t vb_type = get_type_of_register(vb); // interact with type domain if vb_type is ptr_with_offset_t: ptr_with_offset_t vb_type_with_offset = (ptr_with_offset_t)vb_type; int load_at = vb_type_with_offset.offset + vc; if vb_type_with_offset.region is T_CTX: if load_at not in ctx_dists: state.m_reg_dists -= va; // forget the offset of va, as we do not know anything about it else: state.m_reg_dists[va] = ctx_dists[load_at]; else if vb_type_with_offset.region is T_STACK: if load_at not in state.m_stack_dists: state.m_reg_dists -= va; else: state.m_reg_dists[va] = state.m_stack_dists[load_at]; else: state.m_reg_dists -= va; // case when base register type is packet or shared pointer. we do not store any pointers there ``` ``` *(T *)(va arithm_oprtr vb) = vc -> ptr_t va_type = get_type_of_register(va); ptr_t? vc_type = get_type_of_register(vc); // ptr_t or null if vc_type is null: return; // vc is a number if va_type is ptr_with_offset_t: ptr_with_offset_t va_type_with_offset = (ptr_with_offset_t)va_type; int store_at = va_type_with_offset.offset+vb; if va_type_with_offset.region is T_STACK: if vc_type is null: state.m_stack_dists -= store_at; else: state.m_stack_dists[store_at] = state.m_reg_dists[vc]; else: // type is ctx pointer, we do not store any pointers into ctx else: // type is either shared or packet pointer, we do not store any pointers into packet or shared region ``` ``` va = vb -> ptr_t? vb_type = get_type_of_register(vb); if vb_type is null: state.m_reg_dists -= va; else: state.m_reg_dists[va] = state.m_reg_dists[vb]; ``` ``` va = vb arithm_oprtr vc -> ptr_t? vb_type = get_type_of_register(vb); // ptr_t or null // vb is a number if vb_type is null: state.m_reg_dists -= va; // rb is a number; forget type of va return; // vb_type is packet pointer if vb_type.region is packet: dist_t vb_dist = state.m_reg_dists[vb]; if vb_dist is constant_t: weight_t d = vb_dist.dist+vc; state.m_reg_dists[va] = dist_t{slack: vb_dist.slack, dist: d}; else: // not handling the case with non-numeric offsets // all other cases else: state.m_reg_dists -= va; // vb is stack/ctx/shared pointer; forget type of va return; ``` ``` assume va <= vb -> if va not in state.m_reg_dists or vb not in state.m_reg_dists: return; dist_t va_dist = state.m_reg_dists[va]; dist_t vb_dist = state.m_reg_dists[vb]; slack_var_t s = s+string(global_slack_counter++); dist_t f = dist_t{slack: s, dist: va_dist.dist}; dist_t b = dist_t{slack: null, dist: vb_dist.dist}; state.m_extra_constraints[0] = forward_and_backward_eq_t{std::make_pair(f, b)}; state.m_extra_constraints[1] = inequality_t{slack: s, rel: R_GE, 0}; ``` ``` states[bb] = state; ``` # An older version of the Offset inference ``` /* language supported (so far): reg ::= unsigned int stmt ::= reg = *(type *)(reg aop const) | *(type *)(reg aop const) = reg | reg = reg | reg = reg aop val | assume reg rop reg; type ::= u64 | u32 | u16 | u8 aop ::= + | - rop ::= > | <= val ::= reg | const */ ``` ``` using region_t = enum {T_PACKET, T_CTX, T_SHARED, T_STACK}; using register_t = unsigned int; // 0-10 values, representing r0-r10 struct { region_t region; } ptr_t; struct : ptr_t { int offset; } ptr_with_offset_t; ``` ``` using relation_t = enum {R_EQ, R_GT, R_LE}; using location_t = enum {REG, STACK, CTX}; using constant = int; // define a domain for constants using weight_t = constant | symbol; // should be constants + symbols using vert_id = unsigned int; struct node_t { vert_id m_node_id; } struct edge_t { node_t m_from; node_t m_to; weight_t m_dist; relation_t m_relation; } using register_offset_t = std::array<node_t, 11>; using stack_offsets_t = std::unordered_map<int, node_t>; using ctx_offsets_t = std::unordered_map<int, node_t>; using out_edges_t = std::unordered_map<node_t, std::vector<edge_t>>; using stores_pointer_t = std::pair<location_t, unsigned int>; // either REG+register_number, STACK+stack_offset, CTX+ctx_offset using node_pointed_by_t = std::unordered_map<node_t, std::vector<stores_pointer_t>>; // locations the node is being pointed by; could be multiple struct offset_graph_t { out_edges_t m_out_edges; // outgoing edges from each node node_pointed_by_t m_to_node_mappings; } ``` ``` struct state_t = { register_offset_t m_curr_reg_offsets; // mapping of registers to nodes stack_offsets_t m_curr_stack_offsets; // mapping of stack offsets to nodes offset_graph_t m_offset_graph; }; map<cfg_node, state_t> states; // node_t(0) represents packet_start, node_t(1) represents packet_end ctx_offsets_t ctx_offsets = map{76: node_t(0), 80: node_t(1)}; state_t state0; state0.m_offset_graph.m_to_node_mappings[node_t(0)].insert(std::pair(CTX, 76)); state0.m_offset_graph.m_to_node_mappings[node_t(1)].insert(std::pair(CTX, 80)); ``` ``` // given two nodes and their distances from their respective start nodes in two graphs, check if distances and mappings to nodes match def look_for_node_with_same_dist(state_t st, node_t curr_node, weight_t curr_dist, node_t to_search_node, node_t to_search_dist, stores_pointer_t p): if curr_dist is not constant or to_search_dist is not constant: return false; if curr_dist == to_search_dist: if p in st.m_offset_graph.m_to_node_mappings[curr_node]: return true; else if curr_dist < to_search_dist: for out in st.m_offset_graph.m_out_edges[curr_node]: if look_for_node_with_same_dist(st, out.m_to, curr_dist+out.m_dist, to_search_edge, p): return true; return false; else: return false; // given a node and distance from start node in one graph, search for a node with same distance in the other graph def search_in_second_graph(state_t st, node_t to_search_node, weight_t to_search_dist, stores_pointer_t p): data_node = node_t(0); for out_edge in st.m_offset_graph.m_out_edges[data_node]: if look_for_node_with_same_dist(st, out_edge.m_to, out_edge.m_dist, to_search_node, to_search_dist, p): return true; return false; def operator||(state_t st1, state_t st2): state_t state = _|_; node_t data_node = node_t(0); node_t data_end_node = node_t(1); // for data_node and data_end_node, we only need to check if node mappings are consistent in both states, no traversals needed for node in std::vector{data_node, data_end_node}: for pointed_by in st1.m_offset_graph.m_to_node_mappings[node]: if pointed_by in st2.m_offset_graph.m_to_node_mappings[node]: state.m_offset_graph.m_to_node_mappings[node].insert(pointed_by); if pointed_by[0] == REG: state.m_curr_reg_offsets[pointed_by[1]] = node; else: state.m_curr_stack_offset[pointed_by[1]] = node; std::vector<node_t> worklist = std::vector{std::pair(data_node, 0)}; // join the two graphs for (node, dist) in worklist: for out_edge in st1.m_offset_graph.m_out_edges[node]: weight_t dist_from_start = out_edge.m_dist+dist; worklist.push_back(std::pair(out_edge.m_to, dist_from_start)); if out_edge.m_relation == R_EQ: // right now handling only equalities state.m_offset_graph.m_out_edges[node] = out_edge; for pointed_by in st1.m_offset_graph.m_to_node_mappings[out_edge.m_to]: if search_in_second_graph(st2, out_node, dist_from_start, pointed_by): state.m_offset_graph.m_to_node_mappings[node].insert(p); if pointed_by[0] == REG: state.m_curr_reg_offsets[pointed_by[1]] = out_edge.m_to; else: state.m_curr_stack_offset[pointed_by[1]] = out_edge.m_to; else: // handle inequalities return state; ``` ``` def join(cfg_node bb): state_t state = _|_; // state is bottom cfg_node[] all_preds = cfg.get_all_predecessors(bb); for i in [0..all_preds.length()]: if i == 0: state = states[all_preds[i]] else: state = state || states[all_preds[i]]; return state; ``` ``` def add_node(node_t curr_node, register_t dest_reg, weight_t dist): if dist is constant: for edge in state.m_offset_graph.m_out_edges[curr_node]: if edge.m_dist is constant and edge.m_relation is R_EQ: if dist < edge.m_dist: node_t dest_node = node_t(global_node_counter++); edge_t e = {m_from: curr_node, m_to: dest_node, m_dist: dist, m_relation: R_EQ}; edge_t e1 = {m_from: dest_node, m_to: edge.m_to, m_dist: edge.m_dist-dist, m_relation: R_EQ}; state.m_offset_graph.m_out_edges[curr_node].insert(e); state.m_offset_graph.m_out_edges[dest_node].insert(e1); state.m_offset_graph.m_out_edges[curr_node].erase(edge); state.m_curr_reg_offsets[dest_reg] = dest_node; state.m_offset_graph.m_to_node_mappings[dest_node].insert(std::pair(REG,dest_reg)); else if dist == edge.m_dist: state.m_curr_reg_offsets[dest_reg] = edge.m_to; state.m_offset_graph.m_to_node_mappings[edge.m_to].insert(std::pair(REG,dest_reg)); else: add_node(m_to, dest_reg, dist-edge.m_dist); else: node_t dest_node = node_t(global_node_counter++); edge_t e = {m_from: curr_node, m_to: dest_node, m_dist: dist, m_relation: R_EQUAL}; ``` ``` for bb in cfg.topological_sort(): vert_id global_node_counter = 2; if cfg.num_of_predecessors(bb) == 1: state = states[cfg.get_predecessor(bb)]; else if cfg.num_of_predecessors(bb) > 1: state = join_at(bb); for instr in bb: match(instr) with: ``` ``` va = *(T *)(vb arithm_oprtr vc) -> ptr_t vb_type = get_type_of_register(vb); // interact with type domain if vb_type is ptr_with_offset_t: ptr_with_offset_t vb_type_with_offset = (ptr_with_offset_t)vb_type; int load_at = vb_type_with_offset.offset + vc; if vb_type_with_offset.region is T_CTX: if load_at not in ctx_offsets: state.m_offset_graph.m_to_nodes_mappings[state.m_curr_reg_offsets[va]].erase(std::pair(REG,va)); state.m_curr_reg_offsets -= va; // forget the offset of va, as we do not know anything about it else: state.m_curr_reg_offsets[va] = ctx_offsets[load_at]; else if vb_type_with_offset.region is T_STACK: if load_at not in state.m_curr_stack_offsets: state.m_offset_graph.m_to_nodes_mappings[state.m_curr_reg_offsets[va]].erase(std::pair(REG,va)); state.m_curr_reg_offsets -= va; else: node_t n = state.m_curr_stack_offsets[load_at]; state.m_curr_reg_offsets[va] = n; state.m_offset_graph.m_to_node_mappings[n].insert(std::pair(REG,va)); else: state.m_offset_graph.m_to_nodes_mappings[state.m_curr_reg_offsets[va]].erase(std::pair(REG,va)); state.m_curr_reg_offsets -= va; // case when base register type is packet or shared pointer. we do not store any pointers there ``` ``` *(T *)(va arithm_oprtr vb) = vc -> ptr_t va_type = get_type_of_register(va); ptr_t? vc_type = get_type_of_register(vc); // ptr_t or null if vc_type is null: return; // vc is a number if va_type is ptr_with_offset_t: ptr_with_offset_t va_type_with_offset = (ptr_with_offset_t)va_type; int store_at = va_type_with_offset.offset+vb; if va_type_with_offset.region is T_STACK: if vc_type is null: state.m_offset_graph.m_to_nodes_mappings[state.m_curr_stack_offsets[store_at]].erase(std::pair(STACK,store_at)); state.m_curr_stack_offsets -= store_at; else: node_t n = state.m_curr_reg_offsets[vc]; state.m_curr_stack_offsets[store_at] = n; state.m_offset_graph.m_to_node_mappings[n].insert(std::pair(STACK,store_at)); else: // type is ctx pointer, we do not store any pointers into ctx else: // type is either shared or packet pointer, we do not store any pointers into packet or shared region ``` ``` va = vb -> ptr_t? vb_type = get_type_of_register(vb); if vb_type is null: state.m_offset_graph.m_to_nodes_mappings[state.m_curr_reg_offsets[va]].erase(std::pair(REG,va)); state.m_curr_reg_offsets -= va; else: node_t n = state.m_curr_reg_offsets[vb]; state.m_curr_reg_offsets[va] = n; state.m_offset_graph.m_to_node_mappings[n].insert(std::pair(REG,va)); ``` ``` va = vb arithm_oprtr vc -> // remember, vc might be a constant or a known/unknown register ptr_t? vb_type = get_type_of_register(vb); // ptr_t or null // vb is a number if vb_type is null: state.m_offset_graph.m_to_nodes_mappings[state.m_curr_reg_offsets[va]].erase(std::pair(REG,va)); state.m_curr_reg_offsets -= va; // rb is a number; forget type of va return; // vb_type is packet pointer if vb_type.region is packet: node_t vb_node = state.m_curr_reg_offsets[vb]; add_node(vb_node, va, vc); // add a new node for va // all other cases else: state.m_offset_graph.m_to_nodes_mappings[state.m_curr_reg_offsets[va]].erase(std::pair(REG,va)); state.m_curr_reg_offsets -= va; // rb is stack/ctx/shared pointer; forget type of va return; ``` ``` assume va <= vb -> if va not in state.m_curr_reg_offsets or vb not in state.m_curr_reg_offsets: return; node_t va_node = state.m_curr_reg_offsets[va]; node_t vb_node = state.m_curr_reg_offsets[vb]; state.m_offset_graph.m_out_edges[va_node].insert({m_from: va_node, m_to: vb_node, m_dist: 0, m_relation: R_LE}); ``` ``` assume va > vb -> if va not in state.m_curr_reg_offsets or vb not in state.m_curr_reg_offsets: return; node_t va_node = state.m_curr_reg_offsets[va]; node_t vb_node = state.m_curr_reg_offsets[vb]; state.m_offset_graph.m_out_edges[va_node].insert({m_from: va_node, m_to: vb_node, m_dist: 0, m_relation: R_GT}); states[bb] = state; ```

    Import from clipboard

    Paste your markdown or webpage here...

    Advanced permission required

    Your current role can only read. Ask the system administrator to acquire write and comment permission.

    This team is disabled

    Sorry, this team is disabled. You can't edit this note.

    This note is locked

    Sorry, only owner can edit this note.

    Reach the limit

    Sorry, you've reached the max length this note can be.
    Please reduce the content or divide it to more notes, thank you!

    Import from Gist

    Import from Snippet

    or

    Export to Snippet

    Are you sure?

    Do you really want to delete this note?
    All users will lose their connection.

    Create a note from template

    Create a note from template

    Oops...
    This template has been removed or transferred.
    Upgrade
    All
    • All
    • Team
    No template.

    Create a template

    Upgrade

    Delete template

    Do you really want to delete this template?
    Turn this template into a regular note and keep its content, versions, and comments.

    This page need refresh

    You have an incompatible client version.
    Refresh to update.
    New version available!
    See releases notes here
    Refresh to enjoy new features.
    Your user state has changed.
    Refresh to load new user state.

    Sign in

    Forgot password

    or

    By clicking below, you agree to our terms of service.

    Sign in via Facebook Sign in via Twitter Sign in via GitHub Sign in via Dropbox Sign in with Wallet
    Wallet ( )
    Connect another wallet

    New to HackMD? Sign up

    Help

    • English
    • 中文
    • Français
    • Deutsch
    • 日本語
    • Español
    • Català
    • Ελληνικά
    • Português
    • italiano
    • Türkçe
    • Русский
    • Nederlands
    • hrvatski jezik
    • język polski
    • Українська
    • हिन्दी
    • svenska
    • Esperanto
    • dansk

    Documents

    Help & Tutorial

    How to use Book mode

    Slide Example

    API Docs

    Edit in VSCode

    Install browser extension

    Contacts

    Feedback

    Discord

    Send us email

    Resources

    Releases

    Pricing

    Blog

    Policy

    Terms

    Privacy

    Cheatsheet

    Syntax Example Reference
    # Header Header 基本排版
    - Unordered List
    • Unordered List
    1. Ordered List
    1. Ordered List
    - [ ] Todo List
    • Todo List
    > Blockquote
    Blockquote
    **Bold font** Bold font
    *Italics font* Italics font
    ~~Strikethrough~~ Strikethrough
    19^th^ 19th
    H~2~O H2O
    ++Inserted text++ Inserted text
    ==Marked text== Marked text
    [link text](https:// "title") Link
    ![image alt](https:// "title") Image
    `Code` Code 在筆記中貼入程式碼
    ```javascript
    var i = 0;
    ```
    var i = 0;
    :smile: :smile: Emoji list
    {%youtube youtube_id %} Externals
    $L^aT_eX$ LaTeX
    :::info
    This is a alert area.
    :::

    This is a alert area.

    Versions and GitHub Sync
    Get Full History Access

    • Edit version name
    • Delete

    revision author avatar     named on  

    More Less

    Note content is identical to the latest version.
    Compare
      Choose a version
      No search result
      Version not found
    Sign in to link this note to GitHub
    Learn more
    This note is not linked with GitHub
     

    Feedback

    Submission failed, please try again

    Thanks for your support.

    On a scale of 0-10, how likely is it that you would recommend HackMD to your friends, family or business associates?

    Please give us some advice and help us improve HackMD.

     

    Thanks for your feedback

    Remove version name

    Do you want to remove this version name and description?

    Transfer ownership

    Transfer to
      Warning: is a public team. If you transfer note to this team, everyone on the web can find and read this note.

        Link with GitHub

        Please authorize HackMD on GitHub
        • Please sign in to GitHub and install the HackMD app on your GitHub repo.
        • HackMD links with GitHub through a GitHub App. You can choose which repo to install our App.
        Learn more  Sign in to GitHub

        Push the note to GitHub Push to GitHub Pull a file from GitHub

          Authorize again
         

        Choose which file to push to

        Select repo
        Refresh Authorize more repos
        Select branch
        Select file
        Select branch
        Choose version(s) to push
        • Save a new version and push
        • Choose from existing versions
        Include title and tags
        Available push count

        Pull from GitHub

         
        File from GitHub
        File from HackMD

        GitHub Link Settings

        File linked

        Linked by
        File path
        Last synced branch
        Available push count

        Danger Zone

        Unlink
        You will no longer receive notification when GitHub file changes after unlink.

        Syncing

        Push failed

        Push successfully