\n\/\/ Note: this code depends on all ptr-to-mem-funcs being same size\n<\/pre>\n<\/p>\n\nOkay, back to our story.\n<\/p>\n
To call through a pointer to a member function, the “this”\npointer is adjusted by the Adjustor, and then the function\nprovided is called. A call through\na function pointer might be compiled like this:\n<\/p>\n
\nvoid (Derived::*pfn)();\nDerived d;\n(d.*pfn)();\n lea ecx, d ; ecx = \"this\"\n add ecx, pfn[4] ; add adjustor\n call pfn[0] ; call\n<\/pre>\n\nWhen would an adjustor be nonzero? Consider the case above.\nThe function Derived::Base2Method() is really Base2::Base2Method()\nand therefore expects to receive “q” as its “this” pointer.\nIn order to convert a “p” to a “q”, the adjustor must have the\nvalue sizeof(Base1), so that when the first line of Base2::Base2Method()\nexecutes, it receives the expected “q” as its “this” pointer.\n<\/p>\n
\n“But why not just use a thunk instead of manually adding the adjustor?”\nIn other words, why not just use a simple pointer to a thunk that\ngoes like this:\n<\/p>\n
\nDerived::Base2Method thunk:\n add ecx, sizeof(Base1) ; convert \"p\" to \"q\"\n jmp Base2::Base2Method ; continue\n<\/pre>\n\nand use that thunk as the function pointer?\n<\/p>\n
\nThe reason: Function pointer casts.\n<\/p>\n
\nConsider the following code:<\/p>\n
\nvoid (Base2::*pfnBase2)();\nvoid (Derived::*pfnDerived)();\npfnDerived = pfnBase2;\n mov ecx, pfnBase2 ; ecx = address\n mov pfnDerived[0], ecx\n mov pfnDerived[4], sizeof(Base1) ; adjustor!\n<\/pre>\n\nWe start with a pointer to a member function of Base2,\nwhich is a class that uses only single inheritance, so it\nconsists of just a pointer to the code.\nTo assign it to a pointer to a member function\nof Derived, which uses multiple inheritance, we can re-use\nthe function address, but we now need an adjustor so that\nthe pointer “p” can properly be converted to a “q”.\n<\/p>\n
\nNotice that the code doesn’t know what function pfnBase2\npoints to, so it can’t just replace it with the matching thunk.\nIt would have to generate a thunk at runtime and somehow\nuse its psychic powers to decide when the memory can safely\nbe freed. (This is C++. No garbage collector here.)\n<\/p>\n
\nNotice also that when pfnBase2\ngot cast to a pointer to member function of Derived, its size changed,\nsince it went from a pointer to a function in a class that uses only single\ninheritance to a pointer to a function in a class\nthat uses multiple inheritance.\n<\/p>\n
\nCasting a function pointer can change its size!\n<\/p><\/blockquote>\n
\nI bet that you didn’t know that before reading this entry.\n<\/p>\n
\nThere’s still an awful lot more to this topic,\nbut I’m going to stop here before everybody’s head explodes.\n<\/p>\n
\nExercise<\/b>:\nConsider the class<\/p>\n\nclass Base3 { int b3; void Base3Method(); };\nclass Derived2 : public Base3, public Derived { };\n<\/pre>\nHow would the following code be compiled?<\/p>\n
\nvoid (Derived::*pfnDerived)();\nvoid (Derived2::*pfnDerived2();\npfnDerived2 = pfnDerived;\n<\/pre>\n\nAnswer to appear tomorrow.<\/p>\n","protected":false},"excerpt":{"rendered":"
Pointers to member functions are very strange animals. Warning: The discussion that follows is specific to the way pointers to member functions are implemented by the Microsoft Visual C++ compiler. Other compilers may do things differently. Well, okay, if you only use single inheritance, then pointers to member functions are just a pointer to the […]<\/p>\n","protected":false},"author":1069,"featured_media":111744,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[1],"tags":[25],"class_list":["post-40713","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-oldnewthing","tag-code"],"acf":[],"blog_post_summary":"
Pointers to member functions are very strange animals. Warning: The discussion that follows is specific to the way pointers to member functions are implemented by the Microsoft Visual C++ compiler. Other compilers may do things differently. Well, okay, if you only use single inheritance, then pointers to member functions are just a pointer to the […]<\/p>\n","_links":{"self":[{"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/posts\/40713","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/users\/1069"}],"replies":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/comments?post=40713"}],"version-history":[{"count":0,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/posts\/40713\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/media\/111744"}],"wp:attachment":[{"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/media?parent=40713"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/categories?post=40713"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/tags?post=40713"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}