Hello Sascha, On 12/12/25 10:37 AM, Sascha Hauer wrote: > On Thu, Dec 11, 2025 at 09:48:10PM +0100, Ahmad Fatoum wrote: >> nvvar_save will load the extenral environment before writing it back >> with nv changed, which we means we still end up parsing the environment >> in this case, even if we don't execute init scripts or import nv out of >> it. >> >> Fix this to only parse the environment when we actually loaded it >> before. >> >> Signed-off-by: Ahmad Fatoum <[email protected]> >> --- >> common/environment.c | 7 +++++++ >> common/globalvar.c | 8 +++++++- >> include/globalvar.h | 1 + >> 3 files changed, 15 insertions(+), 1 deletion(-) >> >> diff --git a/common/environment.c b/common/environment.c >> index 0e551c90352e..ec14d0629a14 100644 >> --- a/common/environment.c >> +++ b/common/environment.c >> @@ -453,6 +453,7 @@ int envfs_load(const char *filename, const char *dir, >> unsigned flags) >> int envfd; >> int ret = 0; >> size_t size, rsize; >> + __maybe_unused const char *defenv_path; >> >> #ifdef __BAREBOX__ >> if (!IS_ALLOWED(SCONFIG_ENVIRONMENT_LOAD)) >> @@ -531,6 +532,12 @@ int envfs_load(const char *filename, const char *dir, >> unsigned flags) >> >> ret = 0; >> >> +#ifdef CONFIG_NVVAR >> + defenv_path = default_environment_path_get(); >> + if (defenv_path && !strcmp(filename, defenv_path)) >> + nv_var_set_persistable(); >> +#endif >> + >> out: >> close(envfd); >> free(buf); >> diff --git a/common/globalvar.c b/common/globalvar.c >> index 77af6733a6a0..1e06fb43775f 100644 >> --- a/common/globalvar.c >> +++ b/common/globalvar.c >> @@ -15,6 +15,7 @@ >> #include <fnmatch.h> >> >> static int nv_dirty; >> +static bool nv_persistable; >> >> struct device global_device = { >> .name = "global", >> @@ -31,6 +32,11 @@ void nv_var_set_clean(void) >> nv_dirty = 0; >> } >> >> +void nv_var_set_persistable(void) >> +{ >> + nv_persistable = true; >> +} >> + >> void globalvar_remove(const char *name) >> { >> struct param_d *p, *tmp; >> @@ -713,7 +719,7 @@ int nvvar_save(void) >> const char *env = default_environment_path_get(); >> int ret = 0; >> #define TMPDIR "/.env.tmp" >> - if (!nv_dirty || !env) >> + if (!nv_dirty || !env || !nv_persistable) >> return 0; > > With this "nv -s" or whatever calls this just silently does nothing. > This doesn't sound like a desired behaviour. At least a message would be > useful. > > What's the purpose of this patch anyway?
In a later commit, we skip envfs_load if autoload_external_env() is disabled. I thought that it's strange for nv -s to still load the external environment to write variables into it. Cheers, Ahmad > > Sascha > -- Pengutronix e.K. | | Steuerwalder Str. 21 | http://www.pengutronix.de/ | 31137 Hildesheim, Germany | Phone: +49-5121-206917-0 | Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |
