| Message ID | patch-v3-1.5-4b320edc933-20210919T084703Z-avarab@gmail.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | repo-settings.c: refactor for clarity, get rid of hacks etc. | expand |
On Sun, Sep 19, 2021 at 10:47:15AM +0200, Ævar Arnfjörð Bjarmason wrote: > Add fatal wrappers for setenv() and unsetenv(). In d7ac12b25d3 (Add > set_git_dir() function, 2007-08-01) we started checking its return > value, and since 48988c4d0c3 (set_git_dir: die when setenv() fails, > 2018-03-30) we've had set_git_dir_1() die if we couldn't set it. > > Let's provide a wrapper for both, this will be useful in many other > places, a subsequent patch will make another use of xsetenv(). Makes sense. > We could make these return "void" (as far as I can tell there's no > other x*() wrappers that needed to make that decision before), > i.e. our "return 0" is only to indicate that we didn't error, which we > would have died on. Let's return "int" instead to be consistent with > the C library function signatures, including for any future code that > expects a pointer to a setenv()-like function. This may be a little over-clever ;). It is cute, but returning an int makes xsetenv a drop-in replacement for setenv. Which is nice, but it makes it all too-easy to take code like: if (setenv(...) < 0) die(_("...")); and replace it with if (xsetenv(...) < 0) which makes the whole conditional redundant, since the wrappers are guaranteed not to return an error. In other words, I like the idea that s/setenv/x&/ causes a compile-time error, and returning an int from these wrappers prevents that from happening. This may be a little too-theoretical, and you're certainly free to disagree, just my $0.02. > I think it would be OK skip the NULL check of the "name" here for the > calls to die_errno(). Almost all of our setenv() callers are taking a > constant string hardcoded in the source as the first argument, and for > the rest we can probably assume they've done the NULL check > themselves. Even if they didn't, modern C libraries are forgiving > about it (e.g. glibc formatting it as "(null)"), on those that aren't, > well, we were about to die anyway. But let's include the check anyway > for good measure. This I think is a good call. I agree in practice that most times we'd be just fine to pass null to printf() (as we have seen from 88617d11f9 (multi-pack-index: fix potential segfault without sub-command, 2021-07-19) ;)). But there's no reason to rely on risky assumptions when it's easy to avoid doing so. > diff --git a/wrapper.c b/wrapper.c > index 7c6586af321..95f989260cd 100644 > --- a/wrapper.c > +++ b/wrapper.c > @@ -145,6 +145,21 @@ void *xcalloc(size_t nmemb, size_t size) > return ret; > } > > +int xsetenv(const char *name, const char *value, int overwrite) > +{ > + if (setenv(name, value, overwrite)) > + die_errno("setenv(%s, '%s', %d) failed", name ? name : "(null)", > + value, overwrite); > + return 0; > +} > + > +int xunsetenv(const char *name) > +{ > + if (!unsetenv(name)) > + die_errno("unsetenv(%s) failed", name ? name : "(null)"); > + return 0; > +} > + For what it's worth, I find these new messages a little wordy. Maybe we should just sticky "could not (un)set %s"? Thanks, Taylor
On Mon, Sep 20 2021, Taylor Blau wrote: > On Sun, Sep 19, 2021 at 10:47:15AM +0200, Ævar Arnfjörð Bjarmason wrote: >> Add fatal wrappers for setenv() and unsetenv(). In d7ac12b25d3 (Add >> set_git_dir() function, 2007-08-01) we started checking its return >> value, and since 48988c4d0c3 (set_git_dir: die when setenv() fails, >> 2018-03-30) we've had set_git_dir_1() die if we couldn't set it. >> >> Let's provide a wrapper for both, this will be useful in many other >> places, a subsequent patch will make another use of xsetenv(). > > Makes sense. > >> We could make these return "void" (as far as I can tell there's no >> other x*() wrappers that needed to make that decision before), >> i.e. our "return 0" is only to indicate that we didn't error, which we >> would have died on. Let's return "int" instead to be consistent with >> the C library function signatures, including for any future code that >> expects a pointer to a setenv()-like function. > > This may be a little over-clever ;). It is cute, but returning an int > makes xsetenv a drop-in replacement for setenv. Which is nice, but it > makes it all too-easy to take code like: > > if (setenv(...) < 0) > die(_("...")); > > and replace it with > > if (xsetenv(...) < 0) > > which makes the whole conditional redundant, since the wrappers are > guaranteed not to return an error. > > In other words, I like the idea that s/setenv/x&/ causes a compile-time > error, and returning an int from these wrappers prevents that from > happening. > > This may be a little too-theoretical, and you're certainly free to > disagree, just my $0.02. I'm fine with doing that in principle, I couldn't find any "x*()" wrappers with different signatures. Yes, having the compiler complain because you used a "void" return value would be nice. I did have some vague notion that this might interact badly with something in compat, i.e. if setenv() or unsetenv() was a fallback, but I can't think now of why that wouldn't work... >> I think it would be OK skip the NULL check of the "name" here for the >> calls to die_errno(). Almost all of our setenv() callers are taking a >> constant string hardcoded in the source as the first argument, and for >> the rest we can probably assume they've done the NULL check >> themselves. Even if they didn't, modern C libraries are forgiving >> about it (e.g. glibc formatting it as "(null)"), on those that aren't, >> well, we were about to die anyway. But let's include the check anyway >> for good measure. > > This I think is a good call. I agree in practice that most times we'd be > just fine to pass null to printf() (as we have seen from 88617d11f9 > (multi-pack-index: fix potential segfault without sub-command, > 2021-07-19) ;)). But there's no reason to rely on risky assumptions when > it's easy to avoid doing so. *nod* >> diff --git a/wrapper.c b/wrapper.c >> index 7c6586af321..95f989260cd 100644 >> --- a/wrapper.c >> +++ b/wrapper.c >> @@ -145,6 +145,21 @@ void *xcalloc(size_t nmemb, size_t size) >> return ret; >> } >> >> +int xsetenv(const char *name, const char *value, int overwrite) >> +{ >> + if (setenv(name, value, overwrite)) >> + die_errno("setenv(%s, '%s', %d) failed", name ? name : "(null)", >> + value, overwrite); >> + return 0; >> +} >> + >> +int xunsetenv(const char *name) >> +{ >> + if (!unsetenv(name)) >> + die_errno("unsetenv(%s) failed", name ? name : "(null)"); >> + return 0; >> +} >> + > > For what it's worth, I find these new messages a little wordy. Maybe > we should just sticky "could not (un)set %s"? Sure, will change it. I think I may have wanted to include the "overwrite" in the setenv() in some way though, so you'd know if it failed because the libc call "really" failed, v.s. we just had an existing value in the environment and didn't set "overwrite". But reading the docs again that won't work, since it'll succeed if it can't write that key, i.e. the caller with overwrite=0 needs to follow-up with a getenv() to see if their value or someone else's was set (or more probably, they don't care, which is why they used overwrite=0). So it doesn't matter either way for xsetenv(), will make it less chatty.
diff --git a/environment.c b/environment.c index d6b22ede7ea..7d8a949285c 100644 --- a/environment.c +++ b/environment.c @@ -330,8 +330,7 @@ char *get_graft_file(struct repository *r) static void set_git_dir_1(const char *path) { - if (setenv(GIT_DIR_ENVIRONMENT, path, 1)) - die(_("could not set GIT_DIR to '%s'"), path); + xsetenv(GIT_DIR_ENVIRONMENT, path, 1); setup_git_env(path); } diff --git a/git-compat-util.h b/git-compat-util.h index b46605300ab..0b0c0305165 100644 --- a/git-compat-util.h +++ b/git-compat-util.h @@ -875,6 +875,8 @@ void *xmemdupz(const void *data, size_t len); char *xstrndup(const char *str, size_t len); void *xrealloc(void *ptr, size_t size); void *xcalloc(size_t nmemb, size_t size); +int xsetenv(const char *name, const char *value, int overwrite); +int xunsetenv(const char *name); void *xmmap(void *start, size_t length, int prot, int flags, int fd, off_t offset); const char *mmap_os_err(void); void *xmmap_gently(void *start, size_t length, int prot, int flags, int fd, off_t offset); diff --git a/wrapper.c b/wrapper.c index 7c6586af321..95f989260cd 100644 --- a/wrapper.c +++ b/wrapper.c @@ -145,6 +145,21 @@ void *xcalloc(size_t nmemb, size_t size) return ret; } +int xsetenv(const char *name, const char *value, int overwrite) +{ + if (setenv(name, value, overwrite)) + die_errno("setenv(%s, '%s', %d) failed", name ? name : "(null)", + value, overwrite); + return 0; +} + +int xunsetenv(const char *name) +{ + if (!unsetenv(name)) + die_errno("unsetenv(%s) failed", name ? name : "(null)"); + return 0; +} + /* * Limit size of IO chunks, because huge chunks only cause pain. OS X * 64-bit is buggy, returning EINVAL if len >= INT_MAX; and even in
Add fatal wrappers for setenv() and unsetenv(). In d7ac12b25d3 (Add set_git_dir() function, 2007-08-01) we started checking its return value, and since 48988c4d0c3 (set_git_dir: die when setenv() fails, 2018-03-30) we've had set_git_dir_1() die if we couldn't set it. Let's provide a wrapper for both, this will be useful in many other places, a subsequent patch will make another use of xsetenv(). The checking of the return value here is over-eager according to setenv(3) and POSIX. It's documented as returning just -1 or 0, so perhaps we should be checking -1 explicitly. Let's just instead die on any non-zero, if our C library is so broken as to return something else than -1 on error (and perhaps not set errno?) the worst we'll do is die with a nonsensical errno value, but we'll want to die in either case. We could make these return "void" (as far as I can tell there's no other x*() wrappers that needed to make that decision before), i.e. our "return 0" is only to indicate that we didn't error, which we would have died on. Let's return "int" instead to be consistent with the C library function signatures, including for any future code that expects a pointer to a setenv()-like function. I think it would be OK skip the NULL check of the "name" here for the calls to die_errno(). Almost all of our setenv() callers are taking a constant string hardcoded in the source as the first argument, and for the rest we can probably assume they've done the NULL check themselves. Even if they didn't, modern C libraries are forgiving about it (e.g. glibc formatting it as "(null)"), on those that aren't, well, we were about to die anyway. But let's include the check anyway for good measure. 1. https://pubs.opengroup.org/onlinepubs/009604499/functions/setenv.html Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> --- environment.c | 3 +-- git-compat-util.h | 2 ++ wrapper.c | 15 +++++++++++++++ 3 files changed, 18 insertions(+), 2 deletions(-)