ynl: avoid leaks in attr override and spec fixes for C
Subsystems (3)
Jakub Kicinski
maintainercover ynl: avoid leaks in attr override and spec fixes for C
The C rt-link work revealed more problems in existing codegen and classic netlink specs.
Patches 1 - 4 fix issues with the codegen. Patches 1 and 2 are pre-requisites for patch 3. Patch 3 fixes leaking memory if user tries to override already set attr. Patch 4 validates attrs in case kernel sends something we don't expect.
Remaining patches fix and align the specs. Patch 5 changes nesting, the rest are naming adjustments.
Jakub Kicinski (8):
tools: ynl-gen: don't declare loop iterator in place
tools: ynl-gen: move local vars after the opening bracket
tools: ynl-gen: individually free previous values on double set
tools: ynl-gen: make sure we validate subtype of array-nest
netlink: specs: rt-link: add an attr layer around alt-ifname
netlink: specs: rtnetlink: attribute naming corrections
netlink: specs: rt-link: adjust mctp attribute naming
netlink: specs: rt-neigh: prefix struct nfmsg members with ndm
Documentation/netlink/specs/rt_link.yaml | 20 +++-- Documentation/netlink/specs/rt_neigh.yaml | 14 ++-- tools/net/ynl/pyynl/ynl_gen_c.py | 96 +++++++++++++++++------ 3 files changed, 92 insertions(+), 38 deletions(-)
Signature · 1 line
2.49.0
Jakub Kicinski
maintainerPATCH 1/8 tools: ynl-gen: don't declare loop iterator in place
The codegen tries to follow the "old" C style and declare loop iterators at the start of the block / function. Only nested request handling breaks this style, so adjust it.
--- tools/net/ynl/pyynl/ynl_gen_c.py | 25 +++++++++++++++++++++---- 1 file changed, 21 insertions(+), 4 deletions(-)
tools/net/ynl/pyynl/ynl_gen_c.py +21 -4
@@ -654,10 +654,10 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S def attr_put(self, ri, var): if self.attr['type'] in scalars: put_type = self.type - ri.cw.p(f"for (unsigned int i = 0; i < {var}->n_{self.c_name}; i++)") + ri.cw.p(f"for (i = 0; i < {var}->n_{self.c_name}; i++)") ri.cw.p(f"ynl_attr_put_{put_type}(nlh, {self.enum_name}, {var}->{self.c_name}[i]);") elif 'type' not in self.attr or self.attr['type'] == 'nest': - ri.cw.p(f"for (unsigned int i = 0; i < {var}->n_{self.c_name}; i++)") + ri.cw.p(f"for (i = 0; i < {var}->n_{self.c_name}; i++)") self._attr_put_line(ri, var, f"{self.nested_render_name}_put(nlh, " + f"{self.enum_name}, &{var}->{self.c_name}[i])") else:
@@ -1644,11 +1644,23 @@ _C_KW = { def put_req_nested(ri, struct): + local_vars = [] + init_lines = [] + + local_vars.append('struct nlattr *nest;') + init_lines.append("nest = ynl_attr_nest_start(nlh, attr_type);") + + for _, arg in struct.member_list(): + if arg.presence_type() == 'count': + local_vars.append('unsigned int i;') + break + put_req_nested_prototype(ri, struct, suffix='') ri.cw.block_start() - ri.cw.write_func_lvar('struct nlattr *nest;') + ri.cw.write_func_lvar(local_vars) - ri.cw.p("nest = ynl_attr_nest_start(nlh, attr_type);") + for line in init_lines: + ri.cw.p(line) for _, arg in struct.member_list(): arg.attr_put(ri, "obj")
@@ -1850,6 +1862,11 @@ _C_KW = { local_vars += ['size_t hdr_len;', 'void *hdr;'] + for _, attr in ri.struct["request"].member_list(): + if attr.presence_type() == 'count': + local_vars += ['unsigned int i;'] + break + print_prototype(ri, direction, terminate=False) ri.cw.block_start() ri.cw.write_func_lvar(local_vars)
Signature · 1 line
2.49.0
Jakub Kicinski
maintainerPATCH 2/8 tools: ynl-gen: move local vars after the opening bracket
In reply to Jakub KicinskiThe "function writing helper" tries to put local variables between prototype and the opening bracket. Clearly wrong, but up until now nothing actually uses it to write local vars so it wasn't noticed.
--- tools/net/ynl/pyynl/ynl_gen_c.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
tools/net/ynl/pyynl/ynl_gen_c.py +1 -1
@@ -1399,9 +1399,9 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S def write_func(self, qual_ret, name, body, args=None, local_vars=None): self.write_func_prot(qual_ret=qual_ret, name=name, args=args) + self.block_start() self.write_func_lvar(local_vars=local_vars) - self.block_start() for line in body: self.p(line) self.block_end()
Signature · 1 line
2.49.0
Jakub Kicinski
maintainerPATCH 3/8 tools: ynl-gen: individually free previous values on double set
In reply to Jakub KicinskiWhen user calls request_attrA_set() multiple times (for the same attribute), and attrA is of type which allocates memory - we try to free the previously associated values. For array types (including multi-attr) we have only freed the array, but the array may have contained pointers.
Refactor the code generation for free attr and reuse the generated lines in setters to flush out the previous state. Since setters are static inlines in the header we need to add forward declarations for the free helpers of pure nested structs. Track which types get used by arrays and include the right forwad declarations.
At least ethtool string set and bit set would not be freed without this. Tho, admittedly, overriding already set attribute twice is likely a very very rare thing to do.
--- tools/net/ynl/pyynl/ynl_gen_c.py | 62 +++++++++++++++++++++++--------- 1 file changed, 45 insertions(+), 17 deletions(-)
tools/net/ynl/pyynl/ynl_gen_c.py +45 -17
@@ -162,9 +162,15 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S def free_needs_iter(self): return False - def free(self, ri, var, ref): + def _free_lines(self, ri, var, ref): if self.is_multi_val() or self.presence_type() == 'len': - ri.cw.p(f'free({var}->{ref}{self.c_name});') + return [f'free({var}->{ref}{self.c_name});'] + return [] + + def free(self, ri, var, ref): + lines = self._free_lines(ri, var, ref) + for line in lines: + ri.cw.p(line) def arg_member(self, ri): member = self._complex_member_type(ri)
@@ -263,6 +269,10 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S var = "req" member = f"{var}->{'.'.join(ref)}" + local_vars = [] + if self.free_needs_iter(): + local_vars += ['unsigned int i;'] + code = [] presence = '' for i in range(0, len(ref)):
@@ -272,6 +282,10 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S if i == len(ref) - 1 and self.presence_type() != 'bit': continue code.append(presence + ' = 1;') + ref_path = '.'.join(ref[:-1]) + if ref_path: + ref_path += '.' + code += self._free_lines(ri, var, ref_path) code += self._setter_lines(ri, member, presence) func_name = f"{op_prefix(ri, direction, deref=deref)}_set_{'_'.join(ref)}"
@@ -279,7 +293,8 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S alloc = bool([x for x in code if 'alloc(' in x]) if free and not alloc: func_name = '__' + func_name - ri.cw.write_func('static inline void', func_name, body=code, + ri.cw.write_func('static inline void', func_name, local_vars=local_vars, + body=code, args=[f'{type_name(ri, direction, deref=deref)} *{var}'] + self.arg_member(ri))
@@ -482,8 +497,7 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S ['unsigned int len;'] def _setter_lines(self, ri, member, presence): - return [f"free({member});", - f"{presence}_len = strlen({self.c_name});", + return [f"{presence}_len = strlen({self.c_name});", f"{member} = malloc({presence}_len + 1);", f'memcpy({member}, {self.c_name}, {presence}_len);', f'{member}[{presence}_len] = 0;']
@@ -536,8 +550,7 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S ['unsigned int len;'] def _setter_lines(self, ri, member, presence): - return [f"free({member});", - f"{presence}_len = len;", + return [f"{presence}_len = len;", f"{member} = malloc({presence}_len);", f'memcpy({member}, {self.c_name}, {presence}_len);']
@@ -574,12 +587,14 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S def _complex_member_type(self, ri): return self.nested_struct_type - def free(self, ri, var, ref): + def _free_lines(self, ri, var, ref): + lines = [] at = '&' if self.is_recursive_for_op(ri): at = '' - ri.cw.p(f'if ({var}->{ref}{self.c_name})') - ri.cw.p(f'{self.nested_render_name}_free({at}{var}->{ref}{self.c_name});') + lines += [f'if ({var}->{ref}{self.c_name})'] + lines += [f'{self.nested_render_name}_free({at}{var}->{ref}{self.c_name});'] + return lines def _attr_typol(self): return f'.type = YNL_PT_NEST, .nest = &{self.nested_render_name}_nest, '
@@ -632,15 +647,19 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S def free_needs_iter(self): return 'type' not in self.attr or self.attr['type'] == 'nest' - def free(self, ri, var, ref): + def _free_lines(self, ri, var, ref): + lines = [] if self.attr['type'] in scalars: - ri.cw.p(f"free({var}->{ref}{self.c_name});") + lines += [f"free({var}->{ref}{self.c_name});"] elif 'type' not in self.attr or self.attr['type'] == 'nest': - ri.cw.p(f"for (i = 0; i < {var}->{ref}n_{self.c_name}; i++)") - ri.cw.p(f'{self.nested_render_name}_free(&{var}->{ref}{self.c_name}[i]);') - ri.cw.p(f"free({var}->{ref}{self.c_name});") + lines += [ + f"for (i = 0; i < {var}->{ref}n_{self.c_name}; i++)", + f'{self.nested_render_name}_free(&{var}->{ref}{self.c_name}[i]);', + f"free({var}->{ref}{self.c_name});", + ] else: raise Exception(f"Free of MultiAttr sub-type {self.attr['type']} not supported yet") + return lines def _attr_policy(self, policy): return self.base_type._attr_policy(policy)
@@ -666,8 +685,7 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S def _setter_lines(self, ri, member, presence): # For multi-attr we have a count, not presence, hack up the presence presence = presence[:-(len('_present.') + len(self.c_name))] + "n_" + self.c_name - return [f"free({member});", - f"{member} = {self.c_name};", + return [f"{member} = {self.c_name};", f"{presence} = n_{self.c_name};"]
@@ -755,6 +773,7 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S self.request = False self.reply = False self.recursive = False + self.in_multi_val = False # used by a MultiAttr or and legacy arrays self.attr_list = [] self.attrs = dict()
@@ -1122,6 +1141,10 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S if attr in rs_members['reply']: self.pure_nested_structs[nested].reply = True + if spec.is_multi_val(): + child = self.pure_nested_structs.get(nested) + child.in_multi_val = True + self._sort_pure_types() # Propagate the request / reply / recursive
@@ -1136,6 +1159,8 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S struct.child_nests.update(child.child_nests) child.request |= struct.request child.reply |= struct.reply + if spec.is_multi_val(): + child.in_multi_val = True if attr_set in struct.child_nests: struct.recursive = True
@@ -2958,6 +2983,9 @@ _C_KW = { for attr_set, struct in parsed.pure_nested_structs.items(): ri = RenderInfo(cw, parsed, args.mode, "", "", attr_set) print_type_full(ri, struct) + if struct.request and struct.in_multi_val: + free_rsp_nested_prototype(ri) + cw.nl() for op_name, op in parsed.ops.items(): cw.p(f"/* ============== {op.enum_name} ============== */")
Signature · 1 line
2.49.0
Jakub Kicinski
maintainerPATCH 4/8 tools: ynl-gen: make sure we validate subtype of array-nest
In reply to Jakub KicinskiArrayNest AKA indexed-array support currently skips inner type validation. We count the attributes and then we parse them, make sure we call validate, too. Otherwise buggy / unexpected kernel response may lead to crashes.
--- tools/net/ynl/pyynl/ynl_gen_c.py | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-)
tools/net/ynl/pyynl/ynl_gen_c.py +5 -2
@@ -714,8 +714,11 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S def _attr_get(self, ri, var): local_vars = ['const struct nlattr *attr2;'] get_lines = [f'attr_{self.c_name} = attr;', - 'ynl_attr_for_each_nested(attr2, attr)', - f'\t{var}->n_{self.c_name}++;'] + 'ynl_attr_for_each_nested(attr2, attr) {', + '\tif (ynl_attr_validate(yarg, attr2))', + '\t\treturn YNL_PARSE_CB_ERROR;', + f'\t{var}->n_{self.c_name}++;', + '}'] return get_lines, None, local_vars
Signature · 1 line
2.49.0
Jakub Kicinski
maintainerPATCH 5/8 netlink: specs: rt-link: add an attr layer around alt-ifname
In reply to Jakub Kicinskialt-ifname attr is directly placed in requests (as an alternative to ifname) but in responses its wrapped up in IFLA_PROP_LIST and only there is may be multi-attr. See rtnl_fill_prop_list().
--- Documentation/netlink/specs/rt_link.yaml | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-)
Documentation/netlink/specs/rt_link.yaml +8 -3
@@ -1113,11 +1113,10 @@ protonum: 0 - name: prop-list type: nest - nested-attributes: link-attrs + nested-attributes: prop-list-link-attrs - name: alt-ifname type: string - multi-attr: true - name: perm-address type: binary
@@ -1163,6 +1162,13 @@ protonum: 0 - name: netns-immutable type: u8 + - + name: prop-list-link-attrs + subset-of: link-attrs + attributes: + - + name: alt-ifname + multi-attr: true - name: af-spec-attrs attributes:
@@ -2453,7 +2459,6 @@ protonum: 0 - min-mtu - max-mtu - prop-list - - alt-ifname - perm-address - proto-down-reason - parent-dev-name
Signature · 1 line
2.49.0
Jakub Kicinski
maintainerPATCH 6/8 netlink: specs: rtnetlink: attribute naming corrections
In reply to Jakub KicinskiSome attribute names diverge in very minor ways from the C names. These are most likely typos, and they prevent the C codegen from working.
--- Documentation/netlink/specs/rt_link.yaml | 6 +++--- Documentation/netlink/specs/rt_neigh.yaml | 2 +- 2 files changed, 4 insertions(+), 4 deletions(-)
Documentation/netlink/specs/rt_link.yaml +3 -3
@@ -1591,7 +1591,7 @@ protonum: 0 name: nf-call-iptables type: u8 - - name: nf-call-ip6-tables + name: nf-call-ip6tables type: u8 - name: nf-call-arptables
@@ -2083,7 +2083,7 @@ protonum: 0 name: id type: u16 - - name: flag + name: flags type: binary struct: ifla-vlan-flags -
@@ -2171,7 +2171,7 @@ protonum: 0 type: binary struct: ifla-cacheinfo - - name: icmp6-stats + name: icmp6stats type: binary struct: ifla-icmp6-stats -
Documentation/netlink/specs/rt_neigh.yaml +1 -1
@@ -189,7 +189,7 @@ protonum: 0 type: binary display-hint: ipv4 - - name: lladr + name: lladdr type: binary display-hint: mac -
Signature · 1 line
2.49.0
Jakub Kicinski
maintainerPATCH 7/8 netlink: specs: rt-link: adjust mctp attribute naming
In reply to Jakub KicinskiMCTP attribute naming is inconsistent. In C we have:
IFLA_MCTP_NET,
IFLA_MCTP_PHYS_BINDING,
^^^^but in YAML:
- mctp-net
- phys-binding
^
no "mctp"
It's unclear whether the "mctp" part of the name is supposed to be a prefix or part of attribute name. Make it a prefix, seems cleaner, even tho technically phys-binding was added later.
--- Documentation/netlink/specs/rt_link.yaml | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
Documentation/netlink/specs/rt_link.yaml +2 -1
@@ -2185,9 +2185,10 @@ protonum: 0 type: u32 - name: mctp-attrs + name-prefix: ifla-mctp- attributes: - - name: mctp-net + name: net type: u32 - name: phys-binding
Signature · 1 line
2.49.0
Jakub Kicinski
maintainerPATCH 8/8 netlink: specs: rt-neigh: prefix struct nfmsg members with ndm
In reply to Jakub KicinskiAttach ndm- to all members of struct nfmsg. We could possibly use name-prefix just for C, but I don't think we have any precedent for using name-prefix on structs, and other rtnetlink sub-specs give full names for fixed header struct members.
--- Documentation/netlink/specs/rt_neigh.yaml | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
Documentation/netlink/specs/rt_neigh.yaml +6 -6
@@ -13,25 +13,25 @@ protonum: 0 type: struct members: - - name: family + name: ndm-family type: u8 - - name: pad + name: ndm-pad type: pad len: 3 - - name: ifindex + name: ndm-ifindex type: s32 - - name: state + name: ndm-state type: u16 enum: nud-state - - name: flags + name: ndm-flags type: u8 enum: ntf-flags - - name: type + name: ndm-type type: u8 enum: rtm-type -
Signature · 1 line
2.49.0
Quoted 11 lines · Jakub Kicinski
The C rt-link work revealed more problems in existing codegen and classic netlink specs.
Patches 1 - 4 fix issues with the codegen. Patches 1 and 2 are pre-requisites for patch 3. Patch 3 fixes leaking memory if user tries to override already set attr. Patch 4 validates attrs in case kernel sends something we don't expect.
Remaining patches fix and align the specs. Patch 5 changes nesting, the rest are naming adjustments.
Quoted 10 lines
Jakub Kicinski (8):
tools: ynl-gen: don't declare loop iterator in place
tools: ynl-gen: move local vars after the opening bracket
tools: ynl-gen: individually free previous values on double set
tools: ynl-gen: make sure we validate subtype of array-nest
netlink: specs: rt-link: add an attr layer around alt-ifname
netlink: specs: rtnetlink: attribute naming corrections
netlink: specs: rt-link: adjust mctp attribute naming
netlink: specs: rt-neigh: prefix struct nfmsg members with ndm
For the series:
Quoted 5 lines
Documentation/netlink/specs/rt_link.yaml | 20 +++-- Documentation/netlink/specs/rt_neigh.yaml | 14 ++-- tools/net/ynl/pyynl/ynl_gen_c.py | 96 +++++++++++++++++------ 3 files changed, 92 insertions(+), 38 deletions(-)
Re: [PATCH net 2/8] tools: ynl-gen: move local vars after the opening bracket
In reply to Jakub KicinskiQuoted 6 lines · Jakub Kicinski
The "function writing helper" tries to put local variables between prototype and the opening bracket. Clearly wrong, but up until now nothing actually uses it to write local vars so it wasn't noticed.
Signed-off-by Jakub Kicinski <[email protected]>
Re: [PATCH net 1/8] tools: ynl-gen: don't declare loop iterator in place
In reply to Jakub KicinskiQuoted 27 lines · Jakub Kicinski
The codegen tries to follow the "old" C style and declare loop iterators at the start of the block / function. Only nested request handling breaks this style, so adjust it.
Signed-off-by Jakub Kicinski <[email protected]>--- tools/net/ynl/pyynl/ynl_gen_c.py | 25 +++++++++++++++++++++---- 1 file changed, 21 insertions(+), 4 deletions(-)tools/net/ynl/pyynl/ynl_gen_c.py +2
-2@@ -654,10 +654,10 @@ from lib import SpecFamily, SpecAttrSet, SpecAttr, SpecOperation, SpecEnumSet, S def attr_put(self, ri, var): if self.attr['type'] in scalars: put_type = self.type - ri.cw.p(f"for (unsigned int i = 0; i < {var}->n_{self.c_name}; i++)") + ri.cw.p(f"for (i = 0; i < {var}->n_{self.c_name}; i++)") ri.cw.p(f"ynl_attr_put_{put_type}(nlh, {self.enum_name}, {var}->{self.c_name}[i]);") elif 'type' not in self.attr or self.attr['type'] == 'nest': - ri.cw.p(f"for (unsigned int i = 0; i < {var}->n_{self.c_name}; i++)") + ri.cw.p(f"for (i = 0; i < {var}->n_{self.c_name}; i++)") self._attr_put_line(ri, var, f"{self.nested_render_name}_put(nlh, " + f"{self.enum_name}, &{var}->{self.c_name}[i])") else:
tools/net/ynl/pyynl/ynl_gen_c.py _C_KW = { · line 1651
Quoted 8 lines
tools/net/ynl/pyynl/ynl_gen_c.py +5
-0@@ -1644,11 +1644,23 @@ _C_KW = { def put_req_nested(ri, struct): + local_vars = [] + init_lines = [] + + local_vars.append('struct nlattr *nest;') + init_lines.append("nest = ynl_attr_nest_start(nlh, attr_type);")
Minor nit: the series uses += ['...'] elsewhere, is it worth being consistent?
Either way,
Re: [PATCH net 3/8] tools: ynl-gen: individually free previous values on double set
In reply to Jakub KicinskiQuoted 12 lines · Jakub Kicinski
When user calls
request_attrA_set()multiple times (for the same attribute), and attrA is of type which allocates memory - we try to free the previously associated values. For array types (including multi-attr) we have only freed the array, but the array may have contained pointers.Refactor the code generation for free attr and reuse the generated lines in setters to flush out the previous state. Since setters are static inlines in the header we need to add forward declarations for the free helpers of pure nested structs. Track which types get used by arrays and include the right forwad declarations.
Quoted 6 lines
At least ethtool string set and bit set would not be freed without this. Tho, admittedly, overriding already set attribute twice is likely a very very rare thing to do.
Fixes be5bea1cc0bf ("net: add basic C code generators for Netlink")Signed-off-by Jakub Kicinski <[email protected]>
Re: [PATCH net 4/8] tools: ynl-gen: make sure we validate subtype of array-nest
In reply to Jakub KicinskiQuoted 7 lines · Jakub Kicinski
ArrayNest AKA indexed-array support currently skips inner type validation. We count the attributes and then we parse them, make sure we call validate, too. Otherwise buggy / unexpected kernel response may lead to crashes.
Fixes be5bea1cc0bf ("net: add basic C code generators for Netlink")Signed-off-by Jakub Kicinski <[email protected]>
Re: [PATCH net 5/8] netlink: specs: rt-link: add an attr layer around alt-ifname
In reply to Jakub KicinskiQuoted 6 lines · Jakub Kicinski
alt-ifname attr is directly placed in requests (as an alternative to ifname) but in responses its wrapped up in
IFLA_PROP_LISTand only there is may be multi-attr. Seertnl_fill_prop_list().Fixes b2f63d904e72 ("doc/netlink: Add spec for rt link messages")Signed-off-by Jakub Kicinski <[email protected]>
Re: [PATCH net 6/8] netlink: specs: rtnetlink: attribute naming corrections
In reply to Jakub KicinskiQuoted 7 lines · Jakub Kicinski
Some attribute names diverge in very minor ways from the C names. These are most likely typos, and they prevent the C codegen from working.
Fixes bc515ed06652 ("netlink: specs: Add a spec for neighbor tables in rtnetlink")Fixes b2f63d904e72 ("doc/netlink: Add spec for rt link messages")Signed-off-by Jakub Kicinski <[email protected]>
Re: [PATCH net 7/8] netlink: specs: rt-link: adjust mctp attribute naming
In reply to Jakub KicinskiQuoted 11 lines · Jakub Kicinski
MCTP attribute naming is inconsistent. In C we have:
IFLA_MCTP_NET, IFLA_MCTP_PHYS_BINDING, ^^^^but in YAML:
- mctp-net
- phys-binding
^
no "mctp"
Quoted 6 lines
It's unclear whether the "mctp" part of the name is supposed to be a prefix or part of attribute name. Make it a prefix, seems cleaner, even tho technically phys-binding was added later.
Fixes b2f63d904e72 ("doc/netlink: Add spec for rt link messages")Signed-off-by Jakub Kicinski <[email protected]>
Re: [PATCH net 8/8] netlink: specs: rt-neigh: prefix struct nfmsg members with ndm
In reply to Jakub KicinskiQuoted 7 lines · Jakub Kicinski
Attach ndm- to all members of struct nfmsg. We could possibly use name-prefix just for C, but I don't think we have any precedent for using name-prefix on structs, and other rtnetlink sub-specs give full names for fixed header struct members.
Fixes bc515ed06652 ("netlink: specs: Add a spec for neighbor tables in rtnetlink")Signed-off-by Jakub Kicinski <[email protected]>
Jakub Kicinski
maintainerRe: [PATCH net 1/8] tools: ynl-gen: don't declare loop iterator in place
In reply to Donald HunterQuoted 9 lines · On Wed, 16 Apr 2025 11:06:39 +0100 Donald Hunter wrote:
tools/net/ynl/pyynl/ynl_gen_c.py _C_KW = { · line 1651
Quoted 6 lines
def
put_req_nested(ri, struct):tools/net/ynl/pyynl/ynl_gen_c.py +5
-0+ local_vars = [] + init_lines = [] + + local_vars.append('struct nlattr *nest;') + init_lines.append("nest = ynl_attr_nest_start(nlh, attr_type);")Minor nit: the series uses += ['...'] elsewhere, is it worth being consistent?
Agreed, it was annoying me too.. but in _multi_parse() I seem to have used .append() in the exact same scenario. Hard to define "consistent" in this code base :(
RE: [PATCH net 1/8] tools: ynl-gen: don't declare loop iterator in place
Quoted 20 lines
-----Original Message-----
From: Jakub Kicinski <[email protected]>
Sent: Wednesday, April 16, 2025 6:45 AM
To: Donald Hunter <[email protected]>
Cc: [email protected]; [email protected]; Dumazet, Eric <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; Keller, Jacob E <[email protected]>
Subject: Re: [PATCH net 1/8] tools: ynl-gen: don't declare loop iterator in placeQuoted 7 lines · On Wed, 16 Apr 2025 11:06:39 +0100 Donald Hunter wrote:
Quoted 6 lines
def
put_req_nested(ri, struct):tools/net/ynl/pyynl/ynl_gen_c.py +5
-0+ local_vars = [] + init_lines = [] + + local_vars.append('struct nlattr *nest;') + init_lines.append("nest = ynl_attr_nest_start(nlh, attr_type);")
Quoted 6 lines
Quoted 2 lines
Minor nit: the series uses += ['...'] elsewhere, is it worth being consistent?
Agreed, it was annoying me too.. but in
_multi_parse()I seem to have used .append() in the exact same scenario. Hard to define "consistent" in this code base :(
+= [ ... ] is more akin to joining two lists, which may technically be less efficient if you're just adding a single element. Of course that probably doesn't matter for something like this.
I'd say its fairly clear either way so not sure how much we gain from consistency vs re-rolling.
Thanks,
Jake
This series was applied to netdev/net.git (main) by Jakub Kicinski <[email protected]>:
Hello:
This series was applied to netdev/net.git (main)
by Jakub Kicinski <[email protected]>:
Quoted 9 lines · On Mon, 14 Apr 2025 14:18:43 -0700 you wrote:
The C rt-link work revealed more problems in existing codegen and classic netlink specs.
Patches 1 - 4 fix issues with the codegen. Patches 1 and 2 are pre-requisites for patch 3. Patch 3 fixes leaking memory if user tries to override already set attr. Patch 4 validates attrs in case kernel sends something we don't expect.
[...]
Here is the summary with links:
- [net,1/8] tools: ynl-gen: don't declare loop iterator in place
https://git.kernel.org/netdev/net/c/4d07bbf2d456 - [net,2/8] tools: ynl-gen: move local vars after the opening bracket https://git.kernel.org/netdev/net/c/dfa464b4a603 - [net,3/8] tools: ynl-gen: individually free previous values on double set https://git.kernel.org/netdev/net/c/ce6cb8113c84 - [net,4/8] tools: ynl-gen: make sure we validate subtype of array-nest https://git.kernel.org/netdev/net/c/57e7dedf2b8c - [net,5/8] netlink: specs: rt-link: add an attr layer around alt-ifname https://git.kernel.org/netdev/net/c/acf4da17dead - [net,6/8] netlink: specs: rtnetlink: attribute naming corrections https://git.kernel.org/netdev/net/c/540201c0ef7e - [net,7/8] netlink: specs: rt-link: adjust mctp attribute naming https://git.kernel.org/netdev/net/c/beb3c5ad8829 - [net,8/8] netlink: specs: rt-neigh: prefix struct nfmsg members with ndm https://git.kernel.org/netdev/net/c/e31f86ee4b9c
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html