++Arnd for awareness. Hi Paul, On Sat, 15 Dec 2018 at 01:12, Paul Burton <paul.burton@xxxxxxxx> wrote: > > Hi Firoz, > > On Fri, Dec 14, 2018 at 03:53:38PM +0530, Firoz Khan wrote: > > This might be I removed below lines of code from asm/unistd.h > > [PATCH v5 2/7] mips: remove unused macros > > > > -#ifdef CONFIG_MIPS32_N32 > > -#define NR_syscalls (__NR_N32_Linux + __NR_N32_Linux_syscalls) > > -#elif defined(CONFIG_64BIT) > > -#define NR_syscalls (__NR_64_Linux + __NR_64_Linux_syscalls) > > -#else > > -#define NR_syscalls (__NR_O32_Linux + __NR_O32_Linux_syscalls) > > -#endif > > - > > > > Do I need to send another version of the patch series or will you add > > those lines of code back? > > I've undone that change in the test-syscalls branch. > > This code does strike me as odd though - what we're defining NR_syscalls > as here is not a count of syscalls but the maximum syscall number, which > in the worse case (the n32 one) is over 6000. The syscall tracing code > then allocates arrays with that enormous number of entries, when in > reality at least the first 4000 (__NR_O32_Linux) entries will be unused. > Hmm, cleanup for another day added to the TODO list... > > Anyway, I'll let some test builds run & if all looks good I'll push this > to mips-next. Sound good to me. Thanks Firoz
