env: add w flags for net config in explicit write mode
PATCH 1/1 env: add w flags for net config in explicit write mode
From: Adrian Freihofer <[email protected]>
In explicit write access mode (CONFIG_ENV_WRITEABLE_LIST) the environment variables listed in NET_FLAGS and NET6_FLAGS are probably intended to be writeable. Therefore add the 'w' flag to these variables.
---
include/env_flags.h | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+)
Azure build (patch is in a bunch of siemens patches sending now to mainline):
https://dev.azure.com/hs0298/hs/_build/results?buildId=202&view=results
include/env_flags.h +18 -0
@@ -56,6 +56,16 @@ enum env_flags_varaccess {#define ETHADDR_FLAGS "eth" ETHADDR_WILDCARD "addr:mo,"#endif #endif +#ifdef CONFIG_ENV_WRITEABLE_LIST +#define NET_FLAGS \ + "ipaddr:iw," \ + "gatewayip:iw," \ + "netmask:iw," \ + "serverip:iw," \ + "nvlan:dw," \ + "vlan:dw," \ + "dnsip:iw," +#else #define NET_FLAGS \ "ipaddr:i," \ "gatewayip:i," \
@@ -64,16 +74,24 @@ enum env_flags_varaccess {"nvlan:d," \"vlan:d," \ "dnsip:i," +#endif #else #define ETHADDR_FLAGS #define NET_FLAGS #endif #ifdef CONFIG_IPV6 +#ifdef CONFIG_ENV_WRITEABLE_LIST +#define NET6_FLAGS \ + "ip6addr:sw," \ + "serverip6:sw," \ + "gatewayip6:sw," +#else #define NET6_FLAGS \ "ip6addr:s," \ "serverip6:s," \ "gatewayip6:s," +#endif #else #define NET6_FLAGS #endif
Signature · 3 lines
2.20.1 base-commit: 6b2d05748cf3cd6ba417a96c00602b0122e10af6
Hello,
Quoted 6 lines · Heiko Schocher
From: Adrian Freihofer <[email protected]>
In explicit write access mode (
CONFIG_ENV_WRITEABLE_LIST) the environment variables listed inNET_FLAGSandNET6_FLAGSare probably intended to be writeable. Therefore add the 'w' flag to these variables.
Users who enable CONFIG_ENV_WRITEABLE_LIST typically want full control over which environment variables are writable. The new default behavior introduced by your patch might not align with what all users expect or want.
Would overriding the flags using CFG_ENV_FLAGS_LIST_STATIC not work in this case?
Quoted 36 lines
include/env_flags.h +15
-0@@ -56,6 +56,16 @@ enum env_flags_varaccess {#define ETHADDR_FLAGS "eth" ETHADDR_WILDCARD "addr:mo,"#endif #endif +#ifdef CONFIG_ENV_WRITEABLE_LIST +#define NET_FLAGS \ + "ipaddr:iw," \ + "gatewayip:iw," \ + "netmask:iw," \ + "serverip:iw," \ + "nvlan:dw," \ + "vlan:dw," \ + "dnsip:iw," +#else #define NET_FLAGS \ "ipaddr:i," \ "gatewayip:i," \@@ -64,16 +74,24 @@ enum env_flags_varaccess {"nvlan:d," \"vlan:d," \ "dnsip:i," +#endif #else #define ETHADDR_FLAGS #define NET_FLAGS #endif #ifdef CONFIG_IPV6 +#ifdef CONFIG_ENV_WRITEABLE_LIST +#define NET6_FLAGS \ + "ip6addr:sw," \ + "serverip6:sw," \
include/env_flags.h enum env_flags_varaccess { · line 97
Quoted 10 lines
include/env_flags.h +3
-0+ "gatewayip6:sw," +#else #define NET6_FLAGS \ "ip6addr:s," \ "serverip6:s," \ "gatewayip6:s," +#endif #else #define NET6_FLAGS #endif
Best regards,
Signature · 3 lines
Benjamin Robin, Bootlin Embedded Linux and Kernel engineering https://bootlin.com
Hello Benjamin,
Quoted 10 lines · Benjamin ROBIN
Hello,
Quoted 6 lines · Heiko Schocher
From: Adrian Freihofer <[email protected]>
In explicit write access mode (
CONFIG_ENV_WRITEABLE_LIST) the environment variables listed inNET_FLAGSandNET6_FLAGSare probably intended to be writeable. Therefore add the 'w' flag to these variables.
Quoted 6 lines
Users who enable
CONFIG_ENV_WRITEABLE_LISTtypically want full control over which environment variables are writable. The new default behavior introduced by your patch might not align with what all users expect or want.Would overriding the flags using
CFG_ENV_FLAGS_LIST_STATICnot work in this case?
Good question, I have to test... or may Adrian has already done such tests...
But looking into code.. if we add this variables in ENV_WRITEABLE_LIST the results would be, that this variables are twice in ENV_FLAGS_LIST_STATIC
include/env_flags.h
87 #define ENV_FLAGS_LIST_STATIC \
88 ETHADDR_FLAGS \ 89 NET_FLAGS \ 90 NET6_FLAGS \ 91 SERIAL_FLAGS \ 92 CFG_ENV_FLAGS_LIST_STATIC
once through NET_FLAGS and once through CFG_ENV_FLAGS_LIST_STATIC
So, that is not good, and I think, code will find the first entry, and parse it -> so it will not work.
Should we instead in case CONFIG_ENV_WRITEABLE_LIST is enabled only set/allow CFG_ENV_FLAGS_LIST_STATIC in ENV_FLAGS_LIST_STATIC ?
bye,
Heiko
Signature · 5 lines
Nabla Software Engineering HRB 40522 Augsburg Phone: +49 821 45592596 E-Mail: [email protected] Geschäftsführer : Stefano Babic
Hello Heiko,
Quoted 6 lines · Heiko Schocher
Quoted 6 lines
Users who enable
CONFIG_ENV_WRITEABLE_LISTtypically want full control over
which environment variables are writable. The new default behavior introduced by your patch might not align with what all users expect or want.
Quoted 6 lines
Quoted 3 lines
Would overriding the flags using
CFG_ENV_FLAGS_LIST_STATICnot work in this
case?Good question, I have to test... or may Adrian has already done such tests...
Please test.
Quoted 5 lines
But looking into code.. if we add this variables in
ENV_WRITEABLE_LISTthe results would be, that this variables are twice inENV_FLAGS_LIST_STATIC
include/env_flags.h
87 #defineENV_FLAGS_LIST_STATIC\
Quoted 7 lines
88 ETHADDR_FLAGS \ 89 NET_FLAGS \ 90 NET6_FLAGS \ 91 SERIAL_FLAGS \ 92 CFG_ENV_FLAGS_LIST_STATIConce through
NET_FLAGSand once throughCFG_ENV_FLAGS_LIST_STATIC
Yes, the variables with the flags would be declared twice.
Quoted 2 lines
So, that is not good, and I think, code will find the first entry, and parse it -> so it will not work.
What leads you to this assumption?
If you examine the code, you'll notice that:
- The env_attr_lookup() function returns the last entry.
- When iterating over the flags using env_attr_walk(), any previous attribute flags are overridden by subsequent flag declarations.
Quoted 2 lines
Should we instead in case
CONFIG_ENV_WRITEABLE_LISTis enabled only set/allowCFG_ENV_FLAGS_LIST_STATICinENV_FLAGS_LIST_STATIC?
I haven't tested this myself, as I currently don't have a test environment available. Please test it, and if it doesn't work, the correct fix would be to ensure that any future flag declaration overrides previous ones.
Signature · 3 lines
Benjamin Robin, Bootlin Embedded Linux and Kernel engineering https://bootlin.com
Hello Benjamin,
Quoted 9 lines · Benjamin ROBIN
Hello Heiko,
Quoted 6 lines · Heiko Schocher
Quoted 6 lines
Users who enable
CONFIG_ENV_WRITEABLE_LISTtypically want full control over
which environment variables are writable. The new default behavior introduced by your patch might not align with what all users expect or want.
Quoted 8 lines
Quoted 6 lines
Quoted 3 lines
Would overriding the flags using
CFG_ENV_FLAGS_LIST_STATICnot work in this
case?Good question, I have to test... or may Adrian has already done such tests...
Please test.
We will do.
Quoted 19 lines
Quoted 5 lines
But looking into code.. if we add this variables in
ENV_WRITEABLE_LISTthe results would be, that this variables are twice inENV_FLAGS_LIST_STATIC
include/env_flags.hQuoted 8 lines
87 #define ENV_FLAGS_LIST_STATIC \ 88 ETHADDR_FLAGS \ 89 NET_FLAGS \ 90 NET6_FLAGS \ 91 SERIAL_FLAGS \ 92 CFG_ENV_FLAGS_LIST_STATIConce through
NET_FLAGSand once throughCFG_ENV_FLAGS_LIST_STATICYes, the variables with the flags would be declared twice.
Quoted 2 lines
So, that is not good, and I think, code will find the first entry, and parse it -> so it will not work.
Quoted 6 lines
What leads you to this assumption?
If you examine the code, you'll notice that:
- Theenv_attr_lookup()function returns the last entry.
- When iterating over the flags usingenv_attr_walk(), any previous attribute flags are overridden by subsequent flag declarations.
Ah, thats the trick! If this is the case, I aggree, and we can simply drop this patch!
Quoted 7 lines
Quoted 3 lines
Should we instead in case
CONFIG_ENV_WRITEABLE_LISTis enabled only set/allowCFG_ENV_FLAGS_LIST_STATICinENV_FLAGS_LIST_STATIC?I haven't tested this myself, as I currently don't have a test environment available. Please test it, and if it doesn't work, the correct fix would be to ensure that any future flag declaration overrides previous ones.
We will test, thanks!
bye,
Heiko
Signature · 5 lines
Nabla Software Engineering HRB 40522 Augsburg Phone: +49 821 45592596 E-Mail: [email protected] Geschäftsführer : Stefano Babic
Hello Benjamin
Quoted 52 lines · Heiko Schocher
Hello Benjamin,
Quoted 9 lines · Benjamin ROBIN
Hello Heiko,
Quoted 6 lines · Heiko Schocher
Quoted 6 lines
Users who enable
CONFIG_ENV_WRITEABLE_LISTtypically want full control over
which environment variables are writable. The new default behavior introduced by your patch might not align with what all users expect or want.Quoted 8 lines
Quoted 6 lines
Quoted 3 lines
Would overriding the flags using
CFG_ENV_FLAGS_LIST_STATICnot work in this
case?Good question, I have to test... or may Adrian has already done such tests...
Please test.
We will do.
Quoted 18 lines
Quoted 12 lines
But looking into code.. if we add this variables in
ENV_WRITEABLE_LISTthe results would be, that this variables are twice inENV_FLAGS_LIST_STATIC
include/env_flags.h
87 #defineENV_FLAGS_LIST_STATIC\
88ETHADDR_FLAGS\
89NET_FLAGS\
90NET6_FLAGS\
91SERIAL_FLAGS\
92CFG_ENV_FLAGS_LIST_STATICQuoted 1 line
once through
NET_FLAGSand once throughCFG_ENV_FLAGS_LIST_STATICYes, the variables with the flags would be declared twice.
Quoted 2 lines
So, that is not good, and I think, code will find the first entry, and parse it -> so it will not work.
Quoted 6 lines
What leads you to this assumption?
If you examine the code, you'll notice that:
- Theenv_attr_lookup()function returns the last entry.
- When iterating over the flags usingenv_attr_walk(), any previous attribute flags are overridden by subsequent flag declarations.Ah, thats the trick! If this is the case, I aggree, and we can simply drop this patch!
Quoted 1 line
Quoted 8 lines
Quoted 6 lines
Quoted 2 lines
Should we instead in case
CONFIG_ENV_WRITEABLE_LISTis enabled only set/allowCFG_ENV_FLAGS_LIST_STATICinENV_FLAGS_LIST_STATIC?I haven't tested this myself, as I currently don't have a test environment available. Please test it, and if it doesn't work, the correct fix would be to ensure that any future flag declaration overrides previous ones.
We will test, thanks!
Thanks for this hint, we tested your suggestion and it works!
So this patch can be ignored, I already changed the state in patchwork to superseded!
Thanks!
bye,
Heiko
Quoted 3 lines
bye,
Heiko
Signature · 5 lines
Nabla Software Engineering HRB 40522 Augsburg Phone: +49 821 45592596 E-Mail: [email protected] Geschäftsführer : Stefano Babic