happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

llama: handle short reads in direct I/O path (#18504)

This commit is contained in:
triplenom 2025-12-31 21:24:43 -05:00 committed by GitHub
parent 4cd162a123
commit 9e10bd2eaf
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 11 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
src/llama-mmap.cpp
View File

@ -240,9 +240,10 @@ struct llama_file::impl {
throw std::runtime_error("unexpectedly reached end of file");
}
} else {
bool successful = false;
while (!successful) {
off_t ret = read(fd, ptr, len);
size_t bytes_read = 0;
while (bytes_read < len) {
const size_t to_read = len - bytes_read;
ssize_t ret = ::read(fd, reinterpret_cast<char *>(ptr) + bytes_read, to_read);
if (ret == -1) {
if (errno == EINTR) {
@ -251,10 +252,16 @@ struct llama_file::impl {
throw std::runtime_error(format("read error: %s", strerror(errno)));
}
if (ret == 0) {
// EOF: allow if this read was only pulling alignment padding past file end
off_t pos = lseek(fd, 0, SEEK_CUR);
if (pos != -1 && (size_t) pos == size) {
std::memset(reinterpret_cast<char *>(ptr) + bytes_read, 0, len - bytes_read);
return;
}
throw std::runtime_error("unexpectedly reached end of file");
}
successful = true;
bytes_read += (size_t) ret;
}
}
}