Bug#647309: haskell-src-exts: FTBFS on hppa
On 2016-09-18, at 7:38 PM, John David Anglin wrote: > On 2016-09-15, at 12:24 PM, Clint Adams wrote: > >> On Thu, Sep 15, 2016 at 10:40:07AM -0400, John David Anglin wrote: >>> It might be possible to force gcc to do more frequent garbage collection at >>> the expense of compile time. Changing >>> optimizations might help. Maybe program could be restructured... >> >> Got a patch in mind? > > Maybe add hppa to the arches listed here in rules: > > ifneq (,$(filter $(DEB_BUILD_ARCH),mips mipsel)) > DEB_SETUP_GHC6_CONFIGURE_ARGS += --ghc-options="-optc--param > -optcggc-min-expand=10" > endif Adding "hppa" to the above rule resulted in a successful build: https://buildd.debian.org/status/fetch.php?pkg=haskell-src-exts=hppa=1.17.1-1%2Bb2=1474365987 However, the build time increased significantly. Dave -- John David Anglin dave.ang...@bell.net
Bug#647309: haskell-src-exts: FTBFS on hppa
On 2016-09-15, at 12:24 PM, Clint Adams wrote: > On Thu, Sep 15, 2016 at 10:40:07AM -0400, John David Anglin wrote: >> It might be possible to force gcc to do more frequent garbage collection at >> the expense of compile time. Changing >> optimizations might help. Maybe program could be restructured... > > Got a patch in mind? Maybe add hppa to the arches listed here in rules: ifneq (,$(filter $(DEB_BUILD_ARCH),mips mipsel)) DEB_SETUP_GHC6_CONFIGURE_ARGS += --ghc-options="-optc--param -optcggc-min-expand=10" endif Dave -- John David Anglin dave.ang...@bell.net
Bug#647309: haskell-src-exts: FTBFS on hppa
On Thu, Sep 15, 2016 at 10:40:07AM -0400, John David Anglin wrote: > It might be possible to force gcc to do more frequent garbage collection at > the expense of compile time. Changing > optimizations might help. Maybe program could be restructured... Got a patch in mind?
Bug#647309: haskell-src-exts: FTBFS on hppa
See: https://buildd.debian.org/status/fetch.php?pkg=haskell-src-exts=hppa=1.17.1-1%2Bb2=1473916863 We need 82 MB of stack to avoid overflowing stack (CONFIG_MAX_STACK_SIZE_MB). The default is 80 MB. This option is only relevant for linux targets where the stack grows up (parisc, metag). Then, we run out of virtual memory instead of stack space. 1.17.1-1 used to build on hppa. Probably, switch to gcc-6 caused issue. Success is marginal on other 32-bit targets (x32 has same problem). It might be possible to force gcc to do more frequent garbage collection at the expense of compile time. Changing optimizations might help. Maybe program could be restructured... Dave -- John David Anglin dave.ang...@bell.net
