Jan Voung
2014-Dec-19 23:56 UTC
[LLVMdev] [Patches][RFC] What to do about bitcode streaming.
Hi Rafael, We will try out your patch and check to see how it will fit. You also talked about "It might be even possible to drop the requirement for the size to be known: Replace the call to AtEndOfStream by just trying to read more and checking if it failed, but that is a bit more than I wanted to do for this." That is to remove some calls to getSize()? Is there any expectation that getPointer() will always a pointer within a contiguous region of memory? That is, getPointer() is non-virtual and always refers to the Data field, but the overriding implementation could still dynamically grow some buffer and change Data to point to a new buffer (could that be "protected" instead?)? *) We do have a fork of the bitcode reader and bitstream reader. That said, we still use the upstream bitcode reader / bitstream reader in the browser w/ PNaCl. This is used for debugging non-guaranteed-to-be-stable temporary copies of apps in the browser ( https://developer.chrome.com/native-client/devguide/devcycle/debugging#debugging-pnacl-pexes-pepper-35-or-later ). That said, the overlapped compile/download is probably not a big deal for the debugging use case. *) It looks like gzip encoded files will have content-length set to the size of the *gzipped* body, not the size of the original bitcode: http://stackoverflow.com/questions/3819280/content-length-when-using-http-compression. So we'll need to be careful and treat that as "unknown", which is a bit unfortunate because gzip helps reduce bandwidth requirements. *) Otherwise, not all HTTP responses include a content-length, for various reasons, but I don't know how common this is, but I can try to ask chromium-dev or some other mailing list and see how common that is. *) The HTTP response can lie, or the server can be buggy, and give the wrong content-length. I think we can try to do the safe thing here, but will have to be careful. ==== Misc review comments: class DataStreamer { -public: - /// Fetch bytes [start-end) from the stream, and write them to the - /// buffer pointed to by buf. Returns the number of bytes actually written. - virtual size_t GetBytes(unsigned char *buf, size_t len) = 0; + uint64_t Size; + const uint8_t *Data; + + virtual bool advanceToPosImpl(uint64_t Pos) = 0; +public: + const uint8_t *getPointer(uint64_t Pos, unsigned Size); + DataStreamer(); virtual ~DataStreamer(); -}; -DataStreamer *getDataFileStreamer(const std::string &Filename, - std::string *Err); + uint64_t getSize() const { return Size; } + void setSize(uint64_t Val) { Size = Val; } + +protected: + // It is valid to access Data[0..MaxPos) + uint64_t MaxPos; + void init(uint64_t Size, const uint8_t *Data); +}; } Would be nice to document why there is an advanceToPosImpl, etc. Otherwise the patch is removing a bunch of code/comment that refer to the streaming use case, so if that use case is no longer known, then I imagine folks would try to simplify things further =) - // At this point, if there are any function bodies, the current bit is - // pointing to the END_BLOCK record after them. Now make sure the rest - // of the bits in the module have been read. - if (NextUnreadBit) - ParseModule(true); Hmm, didn't read too closely, but just looks different from what it used to do... + const uint8_t *OrigBufPtr = BufPtr; // If we have a wrapper header, parse it and ignore the non-bc file contents. // The magic number is 0x0B17C0DE stored in little endian. if (isBitcodeWrapper(BufPtr, BufEnd)) if (SkipBitcodeWrapperHeader(BufPtr, BufEnd, true)) return Error(BitcodeError::InvalidBitcodeWrapperHeader); //... - if (isBitcodeWrapper(buf, buf + 4)) { - const unsigned char *bitcodeStart = buf; - const unsigned char *bitcodeEnd = buf + 16; - SkipBitcodeWrapperHeader(bitcodeStart, bitcodeEnd, false); - Bytes.dropLeadingBytes(bitcodeStart - buf); - Bytes.setKnownObjectSize(bitcodeEnd - bitcodeStart); - } + uint64_t Skip = BufPtr - OrigBufPtr; + uint64_t Size = BufEnd - BufPtr + Skip; + StreamFile->setSize(Size); + Stream.JumpToBit(Skip * 8); Not quite clear to me that this is the same. Do we have a test of llvm-dis w/ files that have a bitcode wrapper + misc data surrounding the true bitcode contents? The old dropLeadingBytes would track how many bytes were skipped so that later, reading address N would mean reading N + BytesSkipped. It also subtracted the skipped bytes from BytesRead, and I haven't looked closely enough to see if there is similar state being tracked anymore. @@ -218,18 +214,13 @@ public: void freeState(); bool canSkipToPos(size_t pos) const { - // pos can be skipped to if it is a valid address or one byte past the end. - return pos == 0 || BitStream->getBitcodeBytes().isValidAddress( - static_cast<uint64_t>(pos - 1)); + size_t Size = BitStream->getSize(); + return pos < Size; } Is the "pos can be skipped ... or one byte past the end" behavior still needed, or was it some artifact of how isValidAddress worked? On Fri Dec 19 2014 at 1:03:30 PM Rafael Espíndola < rafael.espindola at gmail.com> wrote:> On 19 December 2014 at 15:59, JF Bastien <jfb at chromium.org> wrote: > > Hi Rafael, > > > > Would you mind waiting for Derek to come back from vacation to discuss > this? > > We do use this code and could improve how it's used and tested within > LLVM. > > Derek is the best person to discuss this, he'll be back in mid-January. > > What about the known-size patch? Can you check if that would work for you? > > The issue is not so much having it tested. It is having a sane and > easy to understand interface. > > Cheers, > Rafael >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20141219/f8db3e31/attachment.html>
Rafael Espíndola
2014-Dec-24 16:49 UTC
[LLVMdev] [Patches][RFC] What to do about bitcode streaming.
> That is to remove some calls to getSize()? Is there any expectation that > getPointer() will always a pointer within a contiguous region of memory? > That is, getPointer() is non-virtual and always refers to the Data field, > but the overriding implementation could still dynamically grow some buffer > and change Data to point to a new buffer (could that be "protected" > instead?)?I think it would fail. Clang does things like * readRecord (... &Blob); * Save a pointer to Blob. * readRecord (...&Blob) It should be possible to use a list of buffers and add a getPointerImpl virtual if absolutely necessary, but see bellow.> *) It looks like gzip encoded files will have content-length set to the size > of the *gzipped* body, not the size of the original bitcode: > http://stackoverflow.com/questions/3819280/content-length-when-using-http-compression. > So we'll need to be careful and treat that as "unknown", which is a bit > unfortunate because gzip helps reduce bandwidth requirements.You should try removing the need for getSize, but if that fails, this is so pnacl specific, can't you wrap the bitcode in with a header including the size (this would be kept in the PNaCl tree. It would just be another implementation of the DataStream).> Not quite clear to me that this is the same. Do we have a test of llvm-dis > w/ files that have a bitcode wrapper + misc data surrounding the true > bitcode contents?We have tests that have a wrapper.> The old dropLeadingBytes would track how many bytes were skipped so that > later, reading address N would mean reading N + BytesSkipped. It also > subtracted the skipped bytes from BytesRead, and I haven't looked closely > enough to see if there is similar state being tracked anymore.There isn't. The point is precisely that it is not needed.> @@ -218,18 +214,13 @@ public: > void freeState(); > > bool canSkipToPos(size_t pos) const { > - // pos can be skipped to if it is a valid address or one byte past the > end. > - return pos == 0 || BitStream->getBitcodeBytes().isValidAddress( > - static_cast<uint64_t>(pos - 1)); > + size_t Size = BitStream->getSize(); > + return pos < Size; > } > > Is the "pos can be skipped ... or one byte past the end" behavior still > needed, or was it some artifact of how isValidAddress worked?Doesn't show up on check-all or a lto bootstrap. I have looked at this a bit more, and now I have a more serious issue with the current state of trunk. Check the small attached patch. It just changes llc to use lazy reading of function bodies (not even lazy streaming). With that patch, llc fails with verifier check failures or assertions if the verifier is disabled. So, codegen cannot handle not reading all bodies upfront. How can PNaCl be getting any improvements from this? I understand that codegen of a function at a time is something you might want to add a some point, but lazy streaming was added in Feb 6 22:30:29 2012 and we still don't have it. It seems the best thing to do is delete this. It can always come back if it has a clean implementation and is actually useful (even if only at -O0). Cheers, Rafael -------------- next part -------------- A non-text attachment was scrubbed... Name: t.patch Type: text/x-patch Size: 484 bytes Desc: not available URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20141224/fafdb60c/attachment.bin>
Jan Voung
2014-Dec-29 19:01 UTC
[LLVMdev] [Patches][RFC] What to do about bitcode streaming.
On Wed, Dec 24, 2014 at 8:49 AM, Rafael Espíndola < rafael.espindola at gmail.com> wrote:> > That is to remove some calls to getSize()? Is there any expectation that > > getPointer() will always a pointer within a contiguous region of memory? > > That is, getPointer() is non-virtual and always refers to the Data field, > > but the overriding implementation could still dynamically grow some > buffer > > and change Data to point to a new buffer (could that be "protected" > > instead?)? > > I think it would fail. Clang does things like > > * readRecord (... &Blob); > * Save a pointer to Blob. > * readRecord (...&Blob) > > It should be possible to use a list of buffers and add a > getPointerImpl virtual if absolutely necessary, but see bellow. >Ah okay, I forgot to look at what Clang does.> > > *) It looks like gzip encoded files will have content-length set to the > size > > of the *gzipped* body, not the size of the original bitcode: > > > http://stackoverflow.com/questions/3819280/content-length-when-using-http-compression > . > > So we'll need to be careful and treat that as "unknown", which is a bit > > unfortunate because gzip helps reduce bandwidth requirements. > > You should try removing the need for getSize, but if that fails, this > is so pnacl specific, can't you wrap the bitcode in with a header > including the size (this would be kept in the PNaCl tree. It would > just be another implementation of the DataStream). >Sure, a wrapper is another option. Will have to discuss that with Derek and the other PNaCl folks when they get back, though it might just be that taking streaming out of upstream llvm and keeping it in our own reader is fine.> > Not quite clear to me that this is the same. Do we have a test of > llvm-dis > > w/ files that have a bitcode wrapper + misc data surrounding the true > > bitcode contents? > > We have tests that have a wrapper. > > > The old dropLeadingBytes would track how many bytes were skipped so that > > later, reading address N would mean reading N + BytesSkipped. It also > > subtracted the skipped bytes from BytesRead, and I haven't looked closely > > enough to see if there is similar state being tracked anymore. > > There isn't. The point is precisely that it is not needed. >Okay great, that's certainly much simpler then =)> > > @@ -218,18 +214,13 @@ public: > > void freeState(); > > > > bool canSkipToPos(size_t pos) const { > > - // pos can be skipped to if it is a valid address or one byte past > the > > end. > > - return pos == 0 || BitStream->getBitcodeBytes().isValidAddress( > > - static_cast<uint64_t>(pos - 1)); > > + size_t Size = BitStream->getSize(); > > + return pos < Size; > > } > > > > Is the "pos can be skipped ... or one byte past the end" behavior still > > needed, or was it some artifact of how isValidAddress worked? > > Doesn't show up on check-all or a lto bootstrap. > > > I have looked at this a bit more, and now I have a more serious issue > with the current state of trunk. Check the small attached patch. It > just changes llc to use lazy reading of function bodies (not even lazy > streaming). With that patch, llc fails with verifier check failures or > assertions if the verifier is disabled. > > So, codegen cannot handle not reading all bodies upfront. How can > PNaCl be getting any improvements from this? I understand that codegen > of a function at a time is something you might want to add a some > point, but lazy streaming was added in Feb 6 22:30:29 2012 and we > still don't have it. >Here's what I think happens (the PNaCl tree has some additional changes to make it work). When you switch to getLazyIRFileModule, file reading is suspended once a function body block is encountered, so the function bodies are not materialized yet. Something needs to materialize them. Otherwise, were you seeing that the verifier couldn't find an entry block? See the attached patch, which modifies the FPPassManager::runOnModule to materialize functions before running passes on them, as FunctionPassManager::run(Function &F) would have done. For PNaCl we had been using a FunctionPassManager directly, instead of a PassManager/FPPassManager to get a similar effect (see commented out code in the attached patch). In general PNaCl's version of llc has been careful about which ModulePasses run during CodeGen. If ModulePasses touch the function bodies, then it could materialize the function bodies too early for streaming to be useful. So far, we only have a limited set of Module passes that we run outside of a PassManager. We also don't allow blockaddresses in our subset of the IR, which would have also caused some forward-referenced functions to be materialized.> It seems the best thing to do is delete this. It can always come back > if it has a clean implementation and is actually useful (even if only > at -O0). > > Cheers, > Rafael >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20141229/c5090a34/attachment.html> -------------- next part -------------- A non-text attachment was scrubbed... Name: t2.patch Type: text/x-patch Size: 1812 bytes Desc: not available URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20141229/c5090a34/attachment.bin>