diff mbox series

initramfs: fix 'bad variable name' error in gen_initramfs_list.sh

Message ID 20191230132006.7401-1-masahiroy@kernel.org (mailing list archive)
State New, archived
Headers show
Series initramfs: fix 'bad variable name' error in gen_initramfs_list.sh | expand

Commit Message

Masahiro Yamada Dec. 30, 2019, 1:20 p.m. UTC
Prior to commit 858805b336be ("kbuild: add $(BASH) to run scripts with
bash-extension"), this shell script was almost always run by bash since
bash is usually installed on the system by default.

Now, this script is run by sh, which might be a symlink to dash. On such
distros, the following code emits an error:

  local dev=`LC_ALL=C ls -l "${location}"`

You can reproduce the build error, for example by setting
CONFIG_INITRAMFS_SOURCE="/dev".

    GEN     usr/initramfs_data.cpio.gz
  ./usr/gen_initramfs_list.sh: 131: local: 1: bad variable name
  make[1]: *** [usr/Makefile:61: usr/initramfs_data.cpio.gz] Error 2

This is because `LC_ALL=C ls -l "${location}"` contains spaces.
Surrounding it with double-quotes fixes the error.

Fixes: 858805b336be ("kbuild: add $(BASH) to run scripts with bash-extension")
Reported-by: Jory A. Pratt <anarchy@gentoo.org>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
---

 usr/gen_initramfs_list.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Justin Capella Dec. 30, 2019, 3:13 p.m. UTC | #1
I was looking at this, and in theory there are other problems that
could arise from non-escaped characters, or things with leading
hyphens... In general it isn't great to assume ls output will play
nicely with the internal field separator / IFS. I think a better
solution would be to use something like ${foo@Q} and instead of trying
to scrape ls -l, maybe using stat, since it can be asked to print out
the device number.

But I don't think this patch fixes the problem mentioned at all-- I
think the problem is initfs is not a variable name on line 61:

$(call if_changed,initfs)

https://github.com/torvalds/linux/blob/351c8a09b00b5c51c8f58b016fffe51f87e2d820/usr/Makefile#L61

The Makefile and script look like more patches would be needed to fix
mentioned issue, and I'm kind of wondering what the intent behind
lines 31-32 is...

ramfs-input := $(if $(filter-out "",$(CONFIG_INITRAMFS_SOURCE)), \
$(shell echo $(CONFIG_INITRAMFS_SOURCE)),-d)

why filter nothing, why shell echo?

Quoting and/or proper escaping would be needed in many other places,
and I suspect cpio input is always regenerated... The use of sed to
manually escape things and suggesting using \ instead of using the
argument terminator "--"... If weird paths with spaces and stuff are a
requirement for the script needs some overhauling

On Mon, Dec 30, 2019 at 5:21 AM Masahiro Yamada <masahiroy@kernel.org> wrote:
>
> Prior to commit 858805b336be ("kbuild: add $(BASH) to run scripts with
> bash-extension"), this shell script was almost always run by bash since
> bash is usually installed on the system by default.
>
> Now, this script is run by sh, which might be a symlink to dash. On such
> distros, the following code emits an error:
>
>   local dev=`LC_ALL=C ls -l "${location}"`
>
> You can reproduce the build error, for example by setting
> CONFIG_INITRAMFS_SOURCE="/dev".
>
>     GEN     usr/initramfs_data.cpio.gz
>   ./usr/gen_initramfs_list.sh: 131: local: 1: bad variable name
>   make[1]: *** [usr/Makefile:61: usr/initramfs_data.cpio.gz] Error 2
>
> This is because `LC_ALL=C ls -l "${location}"` contains spaces.
> Surrounding it with double-quotes fixes the error.
>
> Fixes: 858805b336be ("kbuild: add $(BASH) to run scripts with bash-extension")
> Reported-by: Jory A. Pratt <anarchy@gentoo.org>
> Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
> ---
>
>  usr/gen_initramfs_list.sh | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/usr/gen_initramfs_list.sh b/usr/gen_initramfs_list.sh
> index 0aad760fcd8c..2bbac73e6477 100755
> --- a/usr/gen_initramfs_list.sh
> +++ b/usr/gen_initramfs_list.sh
> @@ -128,7 +128,7 @@ parse() {
>                         str="${ftype} ${name} ${location} ${str}"
>                         ;;
>                 "nod")
> -                       local dev=`LC_ALL=C ls -l "${location}"`
> +                       local dev="`LC_ALL=C ls -l "${location}"`"
>                         local maj=`field 5 ${dev}`
>                         local min=`field 6 ${dev}`
>                         maj=${maj%,}
> --
> 2.17.1
>
Masahiro Yamada Dec. 31, 2019, 10:27 a.m. UTC | #2
On Tue, Dec 31, 2019 at 12:13 AM Justin Capella <justincapella@gmail.com> wrote:
>
> I was looking at this, and in theory there are other problems that
> could arise from non-escaped characters, or things with leading
> hyphens... In general it isn't great to assume ls output will play
> nicely with the internal field separator / IFS. I think a better
> solution would be to use something like ${foo@Q} and instead of trying
> to scrape ls -l, maybe using stat, since it can be asked to print out
> the device number.

I am not sure if 'stat' is necessarily preferred over 'ls -l'.

Commit a670b0b4aed129dc11b465c1c330bfe9202023e5
says 'stat' is not standardized.

There is some room for argument
how far we should care about the portability, though.


>
> But I don't think this patch fixes the problem mentioned at all-- I
> think the problem is initfs is not a variable name on line 61:

cmd_initramfs is defined at line 46.

Look into scripts/Kbuild.include
if you want to know how if_changed works.


> $(call if_changed,initfs)
>
> https://github.com/torvalds/linux/blob/351c8a09b00b5c51c8f58b016fffe51f87e2d820/usr/Makefile#L61
>
> The Makefile and script look like more patches would be needed to fix
> mentioned issue, and I'm kind of wondering what the intent behind
> lines 31-32 is...
>
> ramfs-input := $(if $(filter-out "",$(CONFIG_INITRAMFS_SOURCE)), \
> $(shell echo $(CONFIG_INITRAMFS_SOURCE)),-d)
>
> why filter nothing, why shell echo?

It does filter "", which is different from nothing.

You need to notice GNU Make handles double-quote (")
as a normal character.
There is no special meaning as it has in shell.

'echo' is used just for ripping off the double-quotes.


>
> Quoting and/or proper escaping would be needed in many other places,
> and I suspect cpio input is always regenerated...

I do not think so.

If the cpio input is regenerated,
the following log is shown.

GEN     usr/initramfs_data.cpio


I do not see it every time.


> The use of sed to
> manually escape things and suggesting using \ instead of using the
> argument terminator "--"... If weird paths with spaces and stuff are a
> requirement for the script needs some overhauling

If you find a problem in particular, a patch is welcome.

But, what you stated above comes from your misunderstanding.




> On Mon, Dec 30, 2019 at 5:21 AM Masahiro Yamada <masahiroy@kernel.org> wrote:
> >
> > Prior to commit 858805b336be ("kbuild: add $(BASH) to run scripts with
> > bash-extension"), this shell script was almost always run by bash since
> > bash is usually installed on the system by default.
> >
> > Now, this script is run by sh, which might be a symlink to dash. On such
> > distros, the following code emits an error:
> >
> >   local dev=`LC_ALL=C ls -l "${location}"`
> >
> > You can reproduce the build error, for example by setting
> > CONFIG_INITRAMFS_SOURCE="/dev".
> >
> >     GEN     usr/initramfs_data.cpio.gz
> >   ./usr/gen_initramfs_list.sh: 131: local: 1: bad variable name
> >   make[1]: *** [usr/Makefile:61: usr/initramfs_data.cpio.gz] Error 2
> >
> > This is because `LC_ALL=C ls -l "${location}"` contains spaces.
> > Surrounding it with double-quotes fixes the error.
> >
> > Fixes: 858805b336be ("kbuild: add $(BASH) to run scripts with bash-extension")
> > Reported-by: Jory A. Pratt <anarchy@gentoo.org>
> > Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
> > ---
> >
> >  usr/gen_initramfs_list.sh | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/usr/gen_initramfs_list.sh b/usr/gen_initramfs_list.sh
> > index 0aad760fcd8c..2bbac73e6477 100755
> > --- a/usr/gen_initramfs_list.sh
> > +++ b/usr/gen_initramfs_list.sh
> > @@ -128,7 +128,7 @@ parse() {
> >                         str="${ftype} ${name} ${location} ${str}"
> >                         ;;
> >                 "nod")
> > -                       local dev=`LC_ALL=C ls -l "${location}"`
> > +                       local dev="`LC_ALL=C ls -l "${location}"`"
> >                         local maj=`field 5 ${dev}`
> >                         local min=`field 6 ${dev}`
> >                         maj=${maj%,}
> > --
> > 2.17.1
> >
Justin Capella Jan. 2, 2020, 6:03 a.m. UTC | #3
Ah, I was wondering where the magic cmd_ prefix was coming from. But I
still think that there is an issue there-- initfs vs initRAMfs

On Tue, Dec 31, 2019 at 2:28 AM Masahiro Yamada <masahiroy@kernel.org> wrote:
>
> On Tue, Dec 31, 2019 at 12:13 AM Justin Capella <justincapella@gmail.com> wrote:
> >
> > I was looking at this, and in theory there are other problems that
> > could arise from non-escaped characters, or things with leading
> > hyphens... In general it isn't great to assume ls output will play
> > nicely with the internal field separator / IFS. I think a better
> > solution would be to use something like ${foo@Q} and instead of trying
> > to scrape ls -l, maybe using stat, since it can be asked to print out
> > the device number.
>
> I am not sure if 'stat' is necessarily preferred over 'ls -l'.
>
> Commit a670b0b4aed129dc11b465c1c330bfe9202023e5
> says 'stat' is not standardized.
>
> There is some room for argument
> how far we should care about the portability, though.
>
>
> >
> > But I don't think this patch fixes the problem mentioned at all-- I
> > think the problem is initfs is not a variable name on line 61:
>
> cmd_initramfs is defined at line 46.
>
> Look into scripts/Kbuild.include
> if you want to know how if_changed works.
>
>
> > $(call if_changed,initfs)
> >
> > https://github.com/torvalds/linux/blob/351c8a09b00b5c51c8f58b016fffe51f87e2d820/usr/Makefile#L61
> >
> > The Makefile and script look like more patches would be needed to fix
> > mentioned issue, and I'm kind of wondering what the intent behind
> > lines 31-32 is...
> >
> > ramfs-input := $(if $(filter-out "",$(CONFIG_INITRAMFS_SOURCE)), \
> > $(shell echo $(CONFIG_INITRAMFS_SOURCE)),-d)
> >
> > why filter nothing, why shell echo?
>
> It does filter "", which is different from nothing.
>
> You need to notice GNU Make handles double-quote (")
> as a normal character.
> There is no special meaning as it has in shell.
>
> 'echo' is used just for ripping off the double-quotes.
>
>
> >
> > Quoting and/or proper escaping would be needed in many other places,
> > and I suspect cpio input is always regenerated...
>
> I do not think so.
>
> If the cpio input is regenerated,
> the following log is shown.
>
> GEN     usr/initramfs_data.cpio
>
>
> I do not see it every time.
>
>
> > The use of sed to
> > manually escape things and suggesting using \ instead of using the
> > argument terminator "--"... If weird paths with spaces and stuff are a
> > requirement for the script needs some overhauling
>
> If you find a problem in particular, a patch is welcome.
>
> But, what you stated above comes from your misunderstanding.
>
>
>
>
> > On Mon, Dec 30, 2019 at 5:21 AM Masahiro Yamada <masahiroy@kernel.org> wrote:
> > >
> > > Prior to commit 858805b336be ("kbuild: add $(BASH) to run scripts with
> > > bash-extension"), this shell script was almost always run by bash since
> > > bash is usually installed on the system by default.
> > >
> > > Now, this script is run by sh, which might be a symlink to dash. On such
> > > distros, the following code emits an error:
> > >
> > >   local dev=`LC_ALL=C ls -l "${location}"`
> > >
> > > You can reproduce the build error, for example by setting
> > > CONFIG_INITRAMFS_SOURCE="/dev".
> > >
> > >     GEN     usr/initramfs_data.cpio.gz
> > >   ./usr/gen_initramfs_list.sh: 131: local: 1: bad variable name
> > >   make[1]: *** [usr/Makefile:61: usr/initramfs_data.cpio.gz] Error 2
> > >
> > > This is because `LC_ALL=C ls -l "${location}"` contains spaces.
> > > Surrounding it with double-quotes fixes the error.
> > >
> > > Fixes: 858805b336be ("kbuild: add $(BASH) to run scripts with bash-extension")
> > > Reported-by: Jory A. Pratt <anarchy@gentoo.org>
> > > Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
> > > ---
> > >
> > >  usr/gen_initramfs_list.sh | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/usr/gen_initramfs_list.sh b/usr/gen_initramfs_list.sh
> > > index 0aad760fcd8c..2bbac73e6477 100755
> > > --- a/usr/gen_initramfs_list.sh
> > > +++ b/usr/gen_initramfs_list.sh
> > > @@ -128,7 +128,7 @@ parse() {
> > >                         str="${ftype} ${name} ${location} ${str}"
> > >                         ;;
> > >                 "nod")
> > > -                       local dev=`LC_ALL=C ls -l "${location}"`
> > > +                       local dev="`LC_ALL=C ls -l "${location}"`"
> > >                         local maj=`field 5 ${dev}`
> > >                         local min=`field 6 ${dev}`
> > >                         maj=${maj%,}
> > > --
> > > 2.17.1
> > >
>
>
>
> --
> Best Regards
> Masahiro Yamada
diff mbox series

Patch

diff --git a/usr/gen_initramfs_list.sh b/usr/gen_initramfs_list.sh
index 0aad760fcd8c..2bbac73e6477 100755
--- a/usr/gen_initramfs_list.sh
+++ b/usr/gen_initramfs_list.sh
@@ -128,7 +128,7 @@  parse() {
 			str="${ftype} ${name} ${location} ${str}"
 			;;
 		"nod")
-			local dev=`LC_ALL=C ls -l "${location}"`
+			local dev="`LC_ALL=C ls -l "${location}"`"
 			local maj=`field 5 ${dev}`
 			local min=`field 6 ${dev}`
 			maj=${maj%,}