search for: modulesummaryanalysis

...m the assembly, and >> invalidating it after optimization, when reading the Module IR via 'opt' in >> the future. >> >> Does this seem like a reasonable proposal to everyone? >> >> >> I would +1 for making this an analysis pass. How about a >> ModuleSummaryAnalysis that knows how to serialize its result into bitcode >> and YAML format? We can also add a module flag to indicate if the module >> should contain module summary or not. >> >> * For thinLTO compilation, bitcode writer runs the analysis pass if the >> module flag is set,...

...y stands (and in the hopes of faithfully representing the conclusions of that discussion): * Calling getBaseObject() on a GlobalAlias whose aliasee is a GlobalIFunc currently returns null. * Calling getBaseObject() on a GlobalIFunc returns its resolver function. * This causes computeAliasSummary in ModuleSummaryAnalysis to crash on a null dereference for an alias-to-ifunc situation. * A GlobalIFunc and its resolver are *not* interchangeable: at the interface level, they have different signatures (conceptually, the IFunc has the same signature of the function pointer that the resolver potentially returns, not of th...

...ng out how to compute the module summary analysis result from the assembly, and invalidating it after optimization, when reading the Module IR via 'opt' in the future. > > Does this seem like a reasonable proposal to everyone? I would +1 for making this an analysis pass. How about a ModuleSummaryAnalysis that knows how to serialize its result into bitcode and YAML format? We can also add a module flag to indicate if the module should contain module summary or not. * For thinLTO compilation, bitcode writer runs the analysis pass if the module flag is set, and emit module summary into bitcode. * Whe...

...> representing the conclusions of that discussion): >> * Calling getBaseObject() on a GlobalAlias whose aliasee is a >> GlobalIFunc currently returns null. >> * Calling getBaseObject() on a GlobalIFunc returns its resolver function. >> * This causes computeAliasSummary in ModuleSummaryAnalysis to crash on >> a null dereference for an alias-to-ifunc situation. >> * A GlobalIFunc and its resolver are *not* interchangeable: at the >> interface level, they have different signatures (conceptually, the >> IFunc has the same signature of the function pointer that the res...

...ute the module summary analysis result from the assembly, and invalidating it after optimization, when reading the Module IR via 'opt' in the future. >> >> Does this seem like a reasonable proposal to everyone? > > I would +1 for making this an analysis pass. How about a ModuleSummaryAnalysis that knows how to serialize its result into bitcode and YAML format? We can also add a module flag to indicate if the module should contain module summary or not. > > * For thinLTO compilation, bitcode writer runs the analysis pass if the module flag is set, and emit module summary into bitc...

...dule summary analysis result from the assembly, and > invalidating it after optimization, when reading the Module IR via 'opt' in > the future. > > Does this seem like a reasonable proposal to everyone? > > > I would +1 for making this an analysis pass. How about a > ModuleSummaryAnalysis that knows how to serialize its result into bitcode > and YAML format? We can also add a module flag to indicate if the module > should contain module summary or not. > > * For thinLTO compilation, bitcode writer runs the analysis pass if the > module flag is set, and emit module sum...

Hi Mehdi, thanks for the comments, responses and a tweaked proposal below. Teresa On Tue, May 1, 2018 at 11:37 AM, Mehdi AMINI <joker.eph at gmail.com> wrote: > Hi, > > My main concern is this one: > > > Currently, I am emitting the summary entries at the end, after the > metadata nodes. Note that the ModuleSummaryIndex is not currently > referenced from the Module,

Hi, My main concern is this one: > Currently, I am emitting the summary entries at the end, after the metadata nodes. Note that the ModuleSummaryIndex is not currently referenced from the Module, and isn’t currently created when parsing the Module IR bitcode (there is a separate derived class for reading the ModuleSummaryIndex from bitcode). This is because they are not currently used at the