netdev - Re: [PATCH v2] net, uapi: remove inclusion of arpa/inet.h

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 29 Mar 2022 16:12:55 -0700
From:   Nick Desaulniers <ndesaulniers@...gle.com>
To:     Jakub Kicinski <kuba@...nel.org>
Cc:     Jon Maloy <jmaloy@...hat.com>, Ying Xue <ying.xue@...driver.com>,
        Arnd Bergmann <arnd@...db.de>,
        Masahiro Yamada <masahiroy@...nel.org>,
        David Howells <dhowells@...hat.com>,
        Nathan Chancellor <nathan@...nel.org>, netdev@...r.kernel.org,
        tipc-discussion@...ts.sourceforge.net,
        linux-kernel@...r.kernel.org, llvm@...ts.linux.dev,
        Linus Torvalds <torvalds@...ux-foundation.org>
Subject: Re: [PATCH v2] net, uapi: remove inclusion of arpa/inet.h

On Tue, Mar 29, 2022 at 4:01 PM Jakub Kicinski <kuba@...nel.org> wrote:
>
> On Tue, 29 Mar 2022 15:39:56 -0700 Nick Desaulniers wrote:
> > Testing out CONFIG_UAPI_HEADER_TEST=y with a prebuilt Bionic sysroot
> > from Android's SDK, I encountered an error:
> >
> >   HDRTEST usr/include/linux/fsi.h
> > In file included from <built-in>:1:
> > In file included from ./usr/include/linux/tipc_config.h:46:
> > prebuilts/ndk/toolchains/llvm/prebuilt/linux-x86_64/sysroot/usr/include/arpa/inet.h:39:1:
> > error: unknown type name 'in_addr_t'
> > in_addr_t inet_addr(const char* __s);
> > ^
> >
> > This is because Bionic has a bug in its inclusion chain. I sent a patch
> > to fix that, but looking closer at include/uapi/linux/tipc_config.h,
> > there's a comment that it includes arpa/inet.h for ntohs;
> > but ntohs is not defined in any UAPI header. For now, reuse the
> > definitions from include/linux/byteorder/generic.h, since the various
> > conversion functions do exist in UAPI headers:
> > include/uapi/linux/byteorder/big_endian.h
> > include/uapi/linux/byteorder/little_endian.h
> >
> > Link: https://android-review.googlesource.com/c/platform/bionic/+/2048127
> > Signed-off-by: Nick Desaulniers <ndesaulniers@...gle.com>
> > ---
> >  include/uapi/linux/tipc_config.h | 32 ++++++++++++++++----------------
> >  1 file changed, 16 insertions(+), 16 deletions(-)
> >
> > diff --git a/include/uapi/linux/tipc_config.h b/include/uapi/linux/tipc_config.h
> > index 4dfc05651c98..2c494b7ae008 100644
> > --- a/include/uapi/linux/tipc_config.h
> > +++ b/include/uapi/linux/tipc_config.h
> > @@ -43,10 +43,6 @@
> >  #include <linux/tipc.h>
> >  #include <asm/byteorder.h>
> >
> > -#ifndef __KERNEL__
> > -#include <arpa/inet.h> /* for ntohs etc. */
> > -#endif
>
> Hm, how do we know no user space depends on this include?

Without the ability to scan all source code in existence, I guess I
can't prove or disprove that either way.

If this is a reference to "thou shall not break userspace," I don't
think that was in reference to UAPI headers, libc's, or inclusion
chains.

Worst case, someone might have to #include <arpa/inet.h> if they were
relying on transitive dependencies from <linux/tipc_config.h>.  I
don't think we should be helping people write bad code with such
transitive dependencies though.

>
> If nobody screams at us we can try, but then it needs to go into -next,
> and net-next is closed ATM, you'll need to repost once the merge window
> is over.

Ack.

>
> >  /*
> >   * Configuration
> >   *
> > @@ -257,6 +253,10 @@ struct tlv_desc {
> >  #define TLV_SPACE(datalen) (TLV_ALIGN(TLV_LENGTH(datalen)))
> >  #define TLV_DATA(tlv) ((void *)((char *)(tlv) + TLV_LENGTH(0)))
> >
> > +#define __htonl(x) __cpu_to_be32(x)
> > +#define __htons(x) __cpu_to_be16(x)
> > +#define __ntohs(x) __be16_to_cpu(x)
> > +
> >  static inline int TLV_OK(const void *tlv, __u16 space)
> >  {
> >       /*
> > @@ -269,33 +269,33 @@ static inline int TLV_OK(const void *tlv, __u16 space)
> >        */
> >
> >       return (space >= TLV_SPACE(0)) &&
> > -             (ntohs(((struct tlv_desc *)tlv)->tlv_len) <= space);
> > +             (__ntohs(((struct tlv_desc *)tlv)->tlv_len) <= space);
>
> Also why add the defines / macros?
> We could switch to __cpu_to_be16() etc. directly, it seems.

Sure, I thought they might be more readable, but whatever you all
prefer.  Will send a v3 once the merge window closes.
-- 
Thanks,
~Nick Desaulniers