public inbox for cygwin-patches@cygwin.com
 help / color / mirror / Atom feed
From: Johannes Schindelin <Johannes.Schindelin@gmx.de>
To: cygwin-patches@cygwin.com
Subject: Re: [PATCH v2 1/2] Allow deriving the current user's home directory via the HOME variable
Date: Wed, 21 Sep 2022 13:58:48 +0200 (CEST)	[thread overview]
Message-ID: <1r1pq0r7-o3s3-so08-o426-296542797q94@tzk.qr> (raw)
In-Reply-To: <20151217202023.GA3507@calimero.vinschen.de>

Hi Corinna,

sorry for the blast from the past, but I am renewing my efforts to
upstream Git for Windows' patches that can be upstreamed.

On Thu, 17 Dec 2015, Corinna Vinschen wrote:

> On Dec 17 19:05, Johannes Schindelin wrote:
> > [...]
> > diff --git a/winsup/cygwin/uinfo.cc b/winsup/cygwin/uinfo.cc
> > index c9b3e09..a5d6270 100644
> > --- a/winsup/cygwin/uinfo.cc
> > +++ b/winsup/cygwin/uinfo.cc
> > [...]
> > +static size_t
> > +fetch_env(LPCWSTR key, char *buf, size_t size)
>            ^^^
>            space
>
> > +{
> > +  WCHAR wbuf[32767];
>
> Ok, there are a couple of problems here.  First, since this buffer
> is a filename buffer, use NT_MAX_PATH from winsup.h as buffer size.
>
> But then again, please avoid allocating 64K buffers on the stack.
> That's what tmp_pathbuf:w_get () is for.

Excellent. I did it exactly as you suggested.

> > +  DWORD max = sizeof wbuf / sizeof *wbuf;
> > +  DWORD len = GetEnvironmentVariableW (key, wbuf, max);
>
> This call to GetEnvironmentVariableW looks gratuitous to me.  Why don't
> you simply call getenv?  It did the entire job already, it avoids the
> requirement for a local buffer, and in case of $HOME it even did the
> Win32->POSIX path conversion.  If there's a really good reason for using
> GetEnvironmentVariableW it begs at least for a longish comment.

My only worry is that `getenv("HOME")` might receive a "Cygwin-ified"
version of the value. That is, `getenv("HOME")` might return something
like `/cygdrive/c/Users/corinna` when we expect it to return
`C:\Users\corinna` instead.

I do not think that the current iteration is resilient against that.

This problem might not be a big issue with Cygwin (I don't think it
automatically converts environment variables that look like paths from
Windows to Unix-style), but it will most likely cause issues with MSYS2
(where we do precisely that with environment variables that look like
paths). Meaning: it will probably take some follow-up work to make this
work correctly, even if it is just to verify that things work when `HOME`
is in Unix-style already while calling into the runtime.

> > +
> > +  if (!len || len >= max)
> > +    return 0;
> > +
> > +  len = sys_wcstombs (buf, size, wbuf, len);
> > +  return len && len < size ? len : 0;
> > +}
> > +
> > +static char *
> > +fetch_home_env (void)
> > +{
> > +  char home[32767];
> > +  size_t max = sizeof home / sizeof *home, len;
> > +
> > +  if (fetch_env (L"HOME", home, max)
> > +      || ((len = fetch_env (L"HOMEDRIVE", home, max))
> > +        && fetch_env (L"HOMEPATH", home + len, max - len))
> > +      || fetch_env (L"USERPROFILE", home, max))
> > +    {
> > +      tmp_pathbuf tp;
> > +      cygwin_conv_path (CCP_WIN_A_TO_POSIX | CCP_ABSOLUTE,
> > +	  home, tp.c_get(), NT_MAX_PATH);
>                        ^^^
>                        space
> > +      return strdup(tp.c_get());
>                      ^^^      ^^^
>                      space......s
>
> Whoa, tp.c_get() twice to access the same space?  That's a dirty trick
> which may puzzle later readers of the code and heavily depends on
> knowing the internals of tmp_pathbuf.  Please use a variable and only
> assign tp.c_get () once.
>
> OTOH, the above's a case for a cygwin_create_path call, rather than
> cygwin_conv_path+strdup.  Also, if there's *really* a good reason to use
> GetEnvironmentVariableW, you should collapse sys_wcstombs+cygwin_conv_path+
> strdup into a single cygwin_create_path (CCP_WIN_W_TO_POSIX, ...).

Right, that `cygwin_create_path()` call nicely avoids all the problems of
my original code.

>
> > [...]
> > @@ -1079,6 +1123,7 @@ cygheap_pwdgrp::get_shell (cyg_ldap *pldap, cygpsid &sid, PCWSTR dom,
> >  	case NSS_SCHEME_FALLBACK:
> >  	  return NULL;
> >  	case NSS_SCHEME_WINDOWS:
> > +	case NSS_SCHEME_ENV:
> >  	  break;
> >  	case NSS_SCHEME_CYGWIN:
> >  	  if (pldap->fetch_ad_account (sid, false, dnsdomain))
>
> You know that I don't exactly like the "env" idea, but if we implement
> it anyway, wouldn't it make sense to add some kind of $SHELL handling as
> well, for symmetry?

I have decided against that.

The reason: the home directory is a very different thing from the `SHELL`
variable because Windows users _do_ have a home directory even if it is
called differently in Windows speak, while they do not have any POSIX
shell available. There is `COMSPEC`, of course, but it is _not_ a POSIX
shell and cannot be used in place of `SHELL`.

For that reason, I do not believe that we need to do anything about
`SHELL`.

> > [...]
> > @@ -1487,6 +1497,16 @@ of each schema when used with <literal>db_home:</literal>
> >  	      for a detailed description.</listitem>
> >    </varlistentry>
> >    <varlistentry>
> > +    <term><literal>env</literal></term>
> > +    <listitem>Derives the home directory of the current user from the
> > +	      environment variable <literal>HOME</literal> (falling back to
> > +	      <literal>HOMEDRIVE\HOMEPATH</literal> and
> > +	      <literal>USERPROFILE</literal>, in that order).  This is faster
> > +	      than the <term><literal>windows</literal></term> schema at the
> > +	      expense of determining only the current user's home directory
> > +	      correctly.</listitem>
>
> In both case of the documentation it might make sense to add a few words
> along the lines of "This schema is skipped for any other account",
> wouldn't it?

Yes!

(Belated) thank you very much for your review!
Dscho

  reply	other threads:[~2022-09-21 11:58 UTC|newest]

Thread overview: 69+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-09-16 13:06 [PATCH] Allow overriding the " Johannes Schindelin
2015-10-21 18:32 ` Corinna Vinschen
2015-10-22 15:38   ` Johannes Schindelin
2015-10-23  9:10     ` Corinna Vinschen
2015-10-23  9:41       ` Corinna Vinschen
2015-10-23 12:00         ` Johannes Schindelin
2015-12-17 18:05 ` [PATCH v2 0/2] Support deriving the current user's home directory via HOME Johannes Schindelin
2015-12-17 18:05   ` [PATCH v2 1/2] Allow deriving the current user's home directory via the HOME variable Johannes Schindelin
2015-12-17 20:20     ` Corinna Vinschen
2022-09-21 11:58       ` Johannes Schindelin [this message]
2022-10-18 17:02         ` Corinna Vinschen
2022-10-23 21:04           ` Johannes Schindelin
2022-10-24 11:37             ` Corinna Vinschen
2022-11-10 15:16               ` Johannes Schindelin
2022-11-10 15:22                 ` Corinna Vinschen
2022-11-18  8:18                   ` Johannes Schindelin
2022-11-21 11:41                     ` Corinna Vinschen
2023-03-28  8:21                       ` Johannes Schindelin
2015-12-17 18:05   ` [PATCH v2 2/2] Respect `db_home` setting even for the SYSTEM account Johannes Schindelin
2015-12-17 20:49     ` Corinna Vinschen
2015-12-17 21:02       ` Corinna Vinschen
2022-09-21 12:00       ` Johannes Schindelin
2022-09-21 11:51   ` [PATCH v3 0/3] Support deriving the current user's home directory via HOME Johannes Schindelin
2022-09-21 11:51     ` [PATCH v3 1/3] Allow deriving the current user's home directory via the HOME variable Johannes Schindelin
2022-09-21 11:52     ` [PATCH v3 2/3] Respect `db_home` setting even for SYSTEM/Microsoft accounts Johannes Schindelin
2022-09-21 11:52     ` [PATCH v3 3/3] Respect `db_home: env` even when no uid can be determined Johannes Schindelin
2023-03-28  8:17     ` [PATCH v4 0/3] Support deriving the current user's home directory via HOME Johannes Schindelin
2023-03-28  8:17       ` [PATCH v4 1/3] Allow deriving the current user's home directory via the HOME variable Johannes Schindelin
2023-03-28 10:35         ` Corinna Vinschen
2023-03-28 12:34           ` Jon Turney
2023-03-28 13:31             ` Corinna Vinschen
2023-03-29  8:36               ` Corinna Vinschen
2023-04-03  6:39                 ` Johannes Schindelin
2023-03-28  8:17       ` [PATCH v4 2/3] Respect `db_home` setting even for SYSTEM/Microsoft accounts Johannes Schindelin
2023-03-28 10:16         ` Corinna Vinschen
2023-04-03  6:36           ` Johannes Schindelin
2023-04-03 10:59             ` Corinna Vinschen
2023-04-03 13:32               ` Johannes Schindelin
2023-03-28  8:17       ` [PATCH v4 3/3] Respect `db_home: env` even when no uid can be determined Johannes Schindelin
2023-03-28 10:17         ` Corinna Vinschen
2023-04-03  6:45           ` Johannes Schindelin
2023-04-03 13:12             ` Johannes Schindelin
2023-04-03 13:29               ` Corinna Vinschen
2023-04-03 13:57                 ` Johannes Schindelin
2023-04-03 19:23                   ` Corinna Vinschen
2023-04-04 15:11                     ` Johannes Schindelin
2023-04-03 13:19             ` Johannes Schindelin
2023-04-03 14:44       ` [PATCH v5 0/3] Support deriving the current user's home directory via HOME Johannes Schindelin
2023-04-03 14:44         ` [PATCH v5 1/3] Allow deriving the current user's home directory via the HOME variable Johannes Schindelin
2023-04-03 18:36           ` Corinna Vinschen
2023-04-04 15:12             ` Johannes Schindelin
2023-04-03 14:45         ` [PATCH v5 2/3] Respect `db_home` setting even for SYSTEM/Microsoft accounts Johannes Schindelin
2023-04-03 18:37           ` Corinna Vinschen
2023-04-04 15:12             ` Johannes Schindelin
2023-04-03 14:45         ` [PATCH v5 3/3] Respect `db_home: env` even when no uid can be determined Johannes Schindelin
2023-04-04 15:07         ` [PATCH v6 0/4] Support deriving the current user's home directory via HOME Johannes Schindelin
2023-04-04 15:07           ` [PATCH v6 1/4] Allow deriving the current user's home directory via the HOME variable Johannes Schindelin
2023-04-04 15:07           ` [PATCH v6 2/4] Respect `db_home` setting even for SYSTEM/Microsoft accounts Johannes Schindelin
2023-04-04 15:07           ` [PATCH v6 3/4] uinfo: special-case IIS APPPOOL accounts Johannes Schindelin
2023-04-04 15:07           ` [PATCH v6 4/4] Do not rely on `getenv ("HOME")`'s path conversion Johannes Schindelin
2023-04-06  8:37             ` Corinna Vinschen
2023-04-06  9:54               ` Johannes Schindelin
2023-04-06 10:28                 ` Corinna Vinschen
2023-05-22 11:12           ` [PATCH v7 0/4] Support deriving the current user's home directory via HOME Johannes Schindelin
2023-05-22 11:12             ` [PATCH v7 1/4] Allow deriving the current user's home directory via the HOME variable Johannes Schindelin
2023-05-22 11:12             ` [PATCH v7 2/4] Respect `db_home` setting even for SYSTEM/Microsoft accounts Johannes Schindelin
2023-05-22 11:12             ` [PATCH v7 3/4] uinfo: special-case IIS APPPOOL accounts Johannes Schindelin
2023-05-22 11:13             ` [PATCH v7 4/4] Do not rely on `getenv ("HOME")`'s path conversion Johannes Schindelin
2023-06-06 13:33             ` [PATCH v7 0/4] Support deriving the current user's home directory via HOME Corinna Vinschen

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1r1pq0r7-o3s3-so08-o426-296542797q94@tzk.qr \
    --to=johannes.schindelin@gmx.de \
    --cc=cygwin-patches@cygwin.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).