Re: kern/56121: struct kinfo_lwp is inconsistent in NUL-termination of its char arrays, ps assumes all are NUL-terminated!

[David Holland <dholland-bugs%netbsd.org@localhost>]

The following reply was made to PR kern/56121; it has been noted by GNATS.

From: David Holland <dholland-bugs%netbsd.org@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: kern/56121: struct kinfo_lwp is inconsistent in NUL-termination
 of its char arrays, ps assumes all are NUL-terminated!
Date: Sat, 1 May 2021 20:34:46 +0000

 The following response got misfiled by gnats:
 
 From: Simon Burge <simonb%NetBSD.org@localhost>
 To: "Greg A. Woods" <woods%planix.ca@localhost>
 Cc: gnats-bugs%NetBSD.org@localhost
 Subject: Re: struct kinfo_lwp is inconsistent in NUL-termination of its char arrays, ps assumes all are NUL-terminated!
 
  Hi Greg,
 
  "Greg A. Woods" wrote:
 
  > >Number:         56121
  > >Category:       kern
  > >Synopsis:       struct kinfo_lwp is inconsistent in NUL-termination of =
  its char arrays, ps assumes all are NUL-terminated!
 
  > --- sys/sys/sysctl.h.~1.231.~        2021-03-07 17:24:18.000000000 -0800
  > +++ sys/sys/sysctl.h 2021-04-21 17:41:09.797635650 -0700
  > @@ -373,7 +373,7 @@
  >   */
  >  #define     KI_NGROUPS      16
  >  #define     KI_MAXCOMLEN    24      /* extra for 8 byte alignment */
  > -#define     KI_WMESGLEN     8
  > +#define     KI_WMESGLEN     12      /* extra for NUL and 8 byte alignment */
  >  #define     KI_MAXLOGNAME   24      /* extra for 8 byte alignment */
  >  #define     KI_MAXEMULLEN   16
  >  #define     KI_LNAMELEN     20      /* extra 4 for alignment */
  > @@ -566,8 +566,7 @@
  >      uint8_t l_usrpri;               /* U_CHAR: User-priority based on l_cpu and p_nic
  e. */
  >      int8_t  l_stat;                 /* CHAR: S* process status. */
  >      int8_t  l_pad1;                 /* fill out to 4-byte boundary */
  > -    int32_t l_pad2;                 /* .. and then to an 8-byte boundary */
  > -    char    l_wmesg[KI_WMESGLEN];   /* wchan message */
  > +    char    l_wmesg[KI_WMESGLEN];   /* wchan message (with NUL) */
  >      uint64_t l_wchan;               /* PTR: sleep address. */
  >      uint64_t l_cpuid;               /* LONG: CPU id */
  >      uint32_t l_rtime_sec;           /* STRUCT TIMEVAL: Real time. */
 
  This part changes the structure of kinfo_proc2 in a non-backwards
  compatible way by moving the offset of l_wmesg[*].  I think we just need
  to handle the fixed sized buffer directly in ps(1).  I'll look a bit
  more.
 
  Cheers,
  Simon.
 
  [*] For example you should be able to use any ps(1) binary from amd64
      or i386 from NetBSD 1.5 onwards on today's -current amd64 kernel.