| Message ID | 1463442955-8178-1-git-send-email-martin@surround.io |
|---|---|
| State | Superseded |
| Headers | show |
On 05/17/16 01:55, Martin Kelly wrote: > -Og (introduced in GCC 4.8) lets you optimize for debugging experience, > which can be useful for when you want optimized code that is nonetheless > debuggable. > > Signed-off-by: Martin Kelly <martin@surround.io> > --- > Changes based on feedback: > - select --> depends on > - Reworded help text > - Wrapped text to 72 lines Well, actually you didn't: you just copied my text, which I incorrectly wrapped at 78 columns instead of 72... > --- > > Config.in | 10 ++++++++++ > package/Makefile.in | 3 +++ > 2 files changed, 13 insertions(+) > > diff --git a/Config.in b/Config.in > index 9bc8e51..3fe6b7a 100644 > --- a/Config.in > +++ b/Config.in > @@ -510,6 +510,16 @@ config BR2_OPTIMIZE_3 > and also turns on the -finline-functions, -funswitch-loops and > -fgcse-after-reload options. > > +config BR2_OPTIMIZE_g I didn't notice this the first time: config options should be all capitals, like BR2_OPTIMIZE_S (for the -Os option). Regards, Arnout > + bool "optimize for debugging" > + depends on BR2_TOOLCHAIN_GCC_AT_LEAST_4_8 > + help > + Optimize for debugging. This enables optimizations that do not > + interfere with debugging. It should be the optimization level of > + choice for the standard edit-compile-debug cycle, offering a > + reasonable level of optimization while maintaining fast compilation > + and a good debugging experience. > + > config BR2_OPTIMIZE_S > bool "optimize for size" > help > diff --git a/package/Makefile.in b/package/Makefile.in > index 616bdd0..2d6ff89 100644 > --- a/package/Makefile.in > +++ b/package/Makefile.in > @@ -122,6 +122,9 @@ endif > ifeq ($(BR2_OPTIMIZE_3),y) > TARGET_OPTIMIZATION = -O3 > endif > +ifeq ($(BR2_OPTIMIZE_g),y) > +TARGET_OPTIMIZATION = -Og > +endif > ifeq ($(BR2_OPTIMIZE_S),y) > TARGET_OPTIMIZATION = -Os > endif > -- > 2.1.4 > > _______________________________________________ > buildroot mailing list > buildroot@busybox.net > http://lists.busybox.net/mailman/listinfo/buildroot >
On 05/18/2016 12:52 PM, Arnout Vandecappelle wrote: > On 05/17/16 01:55, Martin Kelly wrote: >> -Og (introduced in GCC 4.8) lets you optimize for debugging experience, >> which can be useful for when you want optimized code that is nonetheless >> debuggable. >> >> Signed-off-by: Martin Kelly <martin@surround.io> >> --- >> Changes based on feedback: >> - select --> depends on >> - Reworded help text >> - Wrapped text to 72 lines > > Well, actually you didn't: you just copied my text, which I > incorrectly wrapped at 78 columns instead of 72... > You may have wrapped to 78, but I rewrapped it to 72 columns, so I think the patch I sent is correctly wrapped. >> --- >> >> Config.in | 10 ++++++++++ >> package/Makefile.in | 3 +++ >> 2 files changed, 13 insertions(+) >> >> diff --git a/Config.in b/Config.in >> index 9bc8e51..3fe6b7a 100644 >> --- a/Config.in >> +++ b/Config.in >> @@ -510,6 +510,16 @@ config BR2_OPTIMIZE_3 >> and also turns on the -finline-functions, -funswitch-loops and >> -fgcse-after-reload options. >> >> +config BR2_OPTIMIZE_g > > I didn't notice this the first time: config options should be all > capitals, like BR2_OPTIMIZE_S (for the -Os option). > I will change this and send a revised patch. Note that there are currently several config options that are not all capital (e.g. BR2_STRIP_strip), but perhaps those should change too.
Hello, On Wed, 18 May 2016 13:06:20 -0700, Martin Kelly wrote: > I will change this and send a revised patch. Note that there are > currently several config options that are not all capital (e.g. > BR2_STRIP_strip), but perhaps those should change too. There are indeed previous options that were introduced a long long long time ago, when we didn't had such a thorough review process. Unfortunately, changing them is annoying, as it breaks all the existing .config files for users. For this reason, we try to not rename options just for the beauty of it. Best regards, Thomas
On 05/18/16 22:06, Martin Kelly wrote: > On 05/18/2016 12:52 PM, Arnout Vandecappelle wrote: >> On 05/17/16 01:55, Martin Kelly wrote: >>> -Og (introduced in GCC 4.8) lets you optimize for debugging experience, >>> which can be useful for when you want optimized code that is nonetheless >>> debuggable. >>> >>> Signed-off-by: Martin Kelly <martin@surround.io> >>> --- >>> Changes based on feedback: >>> - select --> depends on >>> - Reworded help text >>> - Wrapped text to 72 lines >> >> Well, actually you didn't: you just copied my text, which I >> incorrectly wrapped at 78 columns instead of 72... >> > > You may have wrapped to 78, but I rewrapped it to 72 columns, so I think the > patch I sent is correctly wrapped. By my count, the line 'reasonable level of optimization while maintaining fast compilation' is 68 characters long. with the tab + 2 spaces that becomes 78. > >>> --- >>> >>> Config.in | 10 ++++++++++ >>> package/Makefile.in | 3 +++ >>> 2 files changed, 13 insertions(+) >>> >>> diff --git a/Config.in b/Config.in >>> index 9bc8e51..3fe6b7a 100644 >>> --- a/Config.in >>> +++ b/Config.in >>> @@ -510,6 +510,16 @@ config BR2_OPTIMIZE_3 >>> and also turns on the -finline-functions, -funswitch-loops and >>> -fgcse-after-reload options. >>> >>> +config BR2_OPTIMIZE_g >> >> I didn't notice this the first time: config options should be all >> capitals, like BR2_OPTIMIZE_S (for the -Os option). >> > > I will change this and send a revised patch. Note that there are currently > several config options that are not all capital (e.g. BR2_STRIP_strip), but > perhaps those should change too. Historical accident. It's not important enough to change it. Regards, Arnout
On 05/18/2016 03:00 PM, Arnout Vandecappelle wrote: > On 05/18/16 22:06, Martin Kelly wrote: >> On 05/18/2016 12:52 PM, Arnout Vandecappelle wrote: >>> On 05/17/16 01:55, Martin Kelly wrote: >>>> -Og (introduced in GCC 4.8) lets you optimize for debugging experience, >>>> which can be useful for when you want optimized code that is >>>> nonetheless >>>> debuggable. >>>> >>>> Signed-off-by: Martin Kelly <martin@surround.io> >>>> --- >>>> Changes based on feedback: >>>> - select --> depends on >>>> - Reworded help text >>>> - Wrapped text to 72 lines >>> >>> Well, actually you didn't: you just copied my text, which I >>> incorrectly wrapped at 78 columns instead of 72... >>> >> >> You may have wrapped to 78, but I rewrapped it to 72 columns, so I >> think the >> patch I sent is correctly wrapped. > > By my count, the line 'reasonable level of optimization while > maintaining fast compilation' is 68 characters long. with the tab + 2 > spaces that becomes 78. > I was working under the assumption that tabs count as 1 character. If tabs count as 8 characters instead, then the rest of the file is already miswrapped; there are many lines containing a tab, 2 spaces, and 70 characters after. Under BR_OPTIMIZE_2, the line starting with "the performance of the generated code" is an example of that. In addition, most text editors seem to count a tab as 1 character when displaying width (Vim certainly does). If the intention is to count a tab as 8 characters, I'd be happy to do so, but then we should also rewrap the rest of the file for consistency. >> >>>> --- >>>> >>>> Config.in | 10 ++++++++++ >>>> package/Makefile.in | 3 +++ >>>> 2 files changed, 13 insertions(+) >>>> >>>> diff --git a/Config.in b/Config.in >>>> index 9bc8e51..3fe6b7a 100644 >>>> --- a/Config.in >>>> +++ b/Config.in >>>> @@ -510,6 +510,16 @@ config BR2_OPTIMIZE_3 >>>> and also turns on the -finline-functions, -funswitch-loops and >>>> -fgcse-after-reload options. >>>> >>>> +config BR2_OPTIMIZE_g >>> >>> I didn't notice this the first time: config options should be all >>> capitals, like BR2_OPTIMIZE_S (for the -Os option). >>> >> >> I will change this and send a revised patch. Note that there are >> currently >> several config options that are not all capital (e.g. >> BR2_STRIP_strip), but >> perhaps those should change too. > > Historical accident. It's not important enough to change it. > Agreed.
On 05/18/2016 03:13 PM, Martin Kelly wrote: > > I was working under the assumption that tabs count as 1 character. If > tabs count as 8 characters instead, then the rest of the file is already > miswrapped; there are many lines containing a tab, 2 spaces, and 70 > characters after. Correction: "there are many lines containing a tab, 2 spaces, and 68 characters after".
On 05/19/16 00:13, Martin Kelly wrote: > On 05/18/2016 03:00 PM, Arnout Vandecappelle wrote: >> On 05/18/16 22:06, Martin Kelly wrote: >>> On 05/18/2016 12:52 PM, Arnout Vandecappelle wrote: >>>> On 05/17/16 01:55, Martin Kelly wrote: >>>>> -Og (introduced in GCC 4.8) lets you optimize for debugging experience, >>>>> which can be useful for when you want optimized code that is >>>>> nonetheless >>>>> debuggable. >>>>> >>>>> Signed-off-by: Martin Kelly <martin@surround.io> >>>>> --- >>>>> Changes based on feedback: >>>>> - select --> depends on >>>>> - Reworded help text >>>>> - Wrapped text to 72 lines >>>> >>>> Well, actually you didn't: you just copied my text, which I >>>> incorrectly wrapped at 78 columns instead of 72... >>>> >>> >>> You may have wrapped to 78, but I rewrapped it to 72 columns, so I >>> think the >>> patch I sent is correctly wrapped. >> >> By my count, the line 'reasonable level of optimization while >> maintaining fast compilation' is 68 characters long. with the tab + 2 >> spaces that becomes 78. >> > > I was working under the assumption that tabs count as 1 character. If tabs count > as 8 characters instead, then the rest of the file is already miswrapped; there > are many lines containing a tab, 2 spaces, and 70 characters after. Under > BR_OPTIMIZE_2, the line starting with "the performance of the generated code" is > an example of that. In addition, most text editors seem to count a tab as 1 > character when displaying width (Vim certainly does). > > If the intention is to count a tab as 8 characters, I'd be happy to do so, but > then we should also rewrap the rest of the file for consistency. You have a point there. I'll Ack your v3, care to send a follow-up that rewraps the entire file? Regards, Arnout [snip]
diff --git a/Config.in b/Config.in index 9bc8e51..3fe6b7a 100644 --- a/Config.in +++ b/Config.in @@ -510,6 +510,16 @@ config BR2_OPTIMIZE_3 and also turns on the -finline-functions, -funswitch-loops and -fgcse-after-reload options. +config BR2_OPTIMIZE_g + bool "optimize for debugging" + depends on BR2_TOOLCHAIN_GCC_AT_LEAST_4_8 + help + Optimize for debugging. This enables optimizations that do not + interfere with debugging. It should be the optimization level of + choice for the standard edit-compile-debug cycle, offering a + reasonable level of optimization while maintaining fast compilation + and a good debugging experience. + config BR2_OPTIMIZE_S bool "optimize for size" help diff --git a/package/Makefile.in b/package/Makefile.in index 616bdd0..2d6ff89 100644 --- a/package/Makefile.in +++ b/package/Makefile.in @@ -122,6 +122,9 @@ endif ifeq ($(BR2_OPTIMIZE_3),y) TARGET_OPTIMIZATION = -O3 endif +ifeq ($(BR2_OPTIMIZE_g),y) +TARGET_OPTIMIZATION = -Og +endif ifeq ($(BR2_OPTIMIZE_S),y) TARGET_OPTIMIZATION = -Os endif
-Og (introduced in GCC 4.8) lets you optimize for debugging experience, which can be useful for when you want optimized code that is nonetheless debuggable. Signed-off-by: Martin Kelly <martin@surround.io> --- Changes based on feedback: - select --> depends on - Reworded help text - Wrapped text to 72 lines --- Config.in | 10 ++++++++++ package/Makefile.in | 3 +++ 2 files changed, 13 insertions(+) -- 2.1.4