1. 20 Oct, 2008 7 commits
  2. 03 Oct, 2008 1 commit
    • Paul Mundt's avatar
      sh: More I/O routine overhauling. · 14866543
      Paul Mundt authored
      This tidies up a lot of the PIO/MMIO split. No in-tree platforms were
      making use of the MMIO overloading through the machvec (nor have any of
      them been in some time), so we just kill all of that off. The ISA I/O
      routine wrapping remains unaffected, which remains the only special
      casing outside of the iomap API that boards need to think about.
      Signed-off-by: default avatarPaul Mundt <lethal@linux-sh.org>
      14866543
  3. 02 Oct, 2008 2 commits
    • Paul Mundt's avatar
      Merge branch 'sh/g3-prep' · bc0f424f
      Paul Mundt authored
      bc0f424f
    • Paul Mundt's avatar
      serial: sh-sci: Dynamic clock management depends on HAVE_CLK. · a2159b52
      Paul Mundt authored
      Presently this is conditionalized on sh, and disabled for sh64.
      Now that SH-5 ties in to the clock framework, the sh64 exception
      can be dropped. Additionally, ARM will want to use the same hooks
      once SH-Mobile G3 grows clock framework support, so switch these
      paths over to HAVE_CLK now.
      
      Once the H8 and ARM sh-sci users hook up HAVE_CLK, the driver can
      be switched over to having an outright dependency on it and the
      ifdefs can go away.
      Signed-off-by: default avatarPaul Mundt <lethal@linux-sh.org>
      a2159b52
  4. 01 Oct, 2008 5 commits
  5. 29 Sep, 2008 5 commits
  6. 24 Sep, 2008 2 commits
  7. 21 Sep, 2008 11 commits
  8. 20 Sep, 2008 2 commits
    • Paul Mundt's avatar
      sh: Support kernel stacks smaller than a page. · c15c5f8c
      Paul Mundt authored
      This follows the powerpc commit f6a61680
      '[POWERPC] Fix kernel stack allocation alignment'.
      
      SH has traditionally forced the thread order to be relative to the page
      size, so there were never any situations where the same bug was
      triggered by slub. Regardless, the usage of > 8kB stacks for the larger
      page sizes is overkill, so we switch to using slab allocations there,
      as per the powerpc change.
      Signed-off-by: default avatarPaul Mundt <lethal@linux-sh.org>
      c15c5f8c
    • Paul Mundt's avatar
      sh: Disable 4kB stacks when using PAGE_SIZE_64KB. · b817f7e0
      Paul Mundt authored
      This combination triggers a divide by zero in kernel/fork.c when
      calculating the initial max_threads value:
      
      	max_threads = mempages / (8 * THREAD_SIZE / PAGE_SIZE);
      
      Simply disable 4K stacks on 64kB PAGE_SIZE to work around this,
      as it's not a terribly useful combination to begin with.
      Signed-off-by: default avatarPaul Mundt <lethal@linux-sh.org>
      b817f7e0
  9. 17 Sep, 2008 5 commits